osiris_phylogenetics: getdata/genbankstrip.pl annotate

annotate getdata/genbankstrip.pl @ 0:5b9a38ec4a39 draft default tip

First commit of old repositories

author	osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
date	Tue, 11 Mar 2014 12:19:13 -0700
parents
children

rev	line source
0 5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1 #!/usr/bin/perl -w
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	2 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	3 # Modified for Osiris by THO.
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	4 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	5 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	6 # GenBankStrip.pl v2.0
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	7 # Last modified July 25, 2005 14:29
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	8 # (c) Olaf R.P. Bininda-Emonds
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	9 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	10 # Input:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	11 # 1) A GenBank output file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	12 # 2) A file containing gene names, each on a separate line
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	13 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	14 # Generates:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	15 # 1) A Se-Al formatted (and optionally a nexus-formatted) datafile containing stripped sequences
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	16 # 2) A summary file giving status of each entry in GenBank file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	17 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	18 # Usage: GenBankStrip.pl -f<filename> [-g<filename>] [-k<number>] [-l<number>] [-o<n\|s>] [-s] [-t] [-h] [-v]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	19 # options: -f<filename> = file containing sequences in GenBank format
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	20 # -g<filename> = file containing specific genes to be stripped
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	21 # -i<genename> = strip a single gene
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	22 # -k<number> = number of (longest) sequences to retain per species for a given gene (default = all)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	23 # -l<number> = minimum length required for all non-tRNA genes (default = none)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	24 # -o<n\|s> = provide output in nexus (n) phytab (p) and/or Se-Al (s) format in addition to fasta format
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	25 # -s = only process sequences with valid species names (i.e., no species with sp. or cf. in names)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	26 # -t<g<number>\|s<number>> = minimum number of sequences (g; default = 0) or species (s; default = 0) a gene must have to be retained; both can be specified simultaneously
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	27 # -h = print this message and quit
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	28 # -v = verbose output
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	29 # -a = Accession number appears first in stripped Fasta File -- Added by THO
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	30 use strict;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	31
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	32 # Set default values
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	33 # Genes and sequences
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	34 my %synonym;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	35 my %complement = ('A' => 'T', 'C' => 'G', 'G' => 'C', 'T' => 'A',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	36 'M' => 'K', 'R' => 'Y', 'W' => 'S', 'S' => 'W', 'Y' => 'R', 'K' => 'M',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	37 'B' => 'V', 'D' => 'H', 'H' => 'D', 'V' => 'B', 'N' => 'N', '-' => '-', '?' => '?');
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	38 my %maxLength;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	39
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	40 # I/O
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	41 my $gbFile;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	42 my ($countFile, $rejectFile, $stripFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	43 my $geneFile;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	44 my %userGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	45 my $singleGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	46 my $keepGene = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	47 my $minLength;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	48 my $minLengthTRNA;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	49 my $speciesBlock = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	50 my $seqThreshold = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	51 my $speciesThreshold = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	52 my $sealPrint = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	53 my $nexusPrint = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	54 my $phytabPrint = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	55
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	56 # Global stats
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	57 my (@globalGeneList, %globalGenePresent, %geneStatus, @rejectList, $stripCount);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	58 my (@speciesList, %speciesPresent, %speciesGenePresent, %quickSpeciesCount, %speciesCount, %quickSequenceCount, %sequenceCount);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	59 $sequenceCount{"all"} = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	60
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	61 # Miscellaneous
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	62 my $debug = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	63 my $verbose = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	64 my $accessionFirst = 0; #Added by THO as option to have accession # first in FASTA output
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	65
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	66 for (my $i = 0; $i <= $#ARGV; $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	67 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	68 #Original stripped out directory structure, required by Galaxy
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	69 # if ($ARGV[$i] =~ /^-f([\w+\.?\w+?]*)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	70 if ($ARGV[$i] =~ /^-f(.+)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	71 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	72 $gbFile = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	73
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	74 # (my $baseFile = $gbFile) =~ s/.\w+$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	75 # $countFile .= $baseFile . "_geneCount.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	76 # $rejectFile .= $baseFile . "_gbs.rejectlist.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	77 # $stripFile .= $baseFile . "_gbs.striplist.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	78 # }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	79 #for Galaxy, easier to have same file name each time it'r run
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	80 (my $baseFile = $gbFile) =~ s/.\w+$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	81 $countFile .= "geneCount.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	82 $rejectFile .= "rejectlist.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	83 $stripFile .= "striplist.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	84 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	85 elsif ($ARGV[$i] =~ /^-g([\w+\.?\w+?]*)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	86 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	87 $geneFile = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	88
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	89 #This is a hack to write the name of a single gene into a new genefile based on the Galaxy call
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	90
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	91 open (GENE,">$geneFile") or die "Cannot open file $geneFile containing gene names.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	92 print GENE "$geneFile\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	93 close GENE;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	94 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	95 elsif ($ARGV[$i] =~ /^-i([\w+\.?\w+?]*)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	96 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	97 $singleGene = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	98 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	99 elsif ($ARGV[$i] =~ /^-k(\d+)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	100 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	101 $keepGene = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	102 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	103 elsif ($ARGV[$i] =~ /^-l(\d+)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	104 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	105 $minLength = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	106 $minLengthTRNA = 50;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	107 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	108 elsif ($ARGV[$i] =~ /^-on/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	109 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	110 $nexusPrint = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	111 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	112 elsif ($ARGV[$i] =~ /^-op/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	113 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	114 $phytabPrint = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	115 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	116 elsif ($ARGV[$i] =~ /^-os/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	117 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	118 $sealPrint = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	119 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	120 elsif ($ARGV[$i] =~ /^-s/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	121 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	122 $speciesBlock = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	123 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	124 elsif ($ARGV[$i] =~ /^-tg(\d+)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	125 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	126 $seqThreshold = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	127 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	128 elsif ($ARGV[$i] =~ /^-ts(\d+)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	129 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	130 $speciesThreshold = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	131 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	132 elsif ($ARGV[$i] =~ /^-v/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	133 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	134 $verbose = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	135 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	136 elsif ($ARGV[$i] =~ /^-a/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	137 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	138 $accessionFirst = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	139 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	140 elsif ($ARGV[$i] =~ /^-x/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	141 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	142 $debug = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	143 $verbose = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	144 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	145 elsif ($ARGV[$i] =~ /^-h/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	146 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	147 print "Usage: GenBankStrip.pl -f<filename> [-g<filename>] [-k<number>] [-l<number>] [-o<n\|s>] [-s] [-t] [-h] [-v]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	148 print "Options: -f<filename> = file containing sequences in GenBank format\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	149 print " -g<filename> = file containing specific genes to be stripped\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	150 print " -k<number> = number of (longest) sequences to retain per species for a given gene (default = all)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	151 print " -l<number> = minimum length required for all non-tRNA genes (default = none)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	152 print " -o<n\|s> = provide output in nexus (n) phytab (p) and/or Se-Al (s) format in addition to fasta format\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	153 print " -s = only process sequences for valid species names (i.e., no species with sp. or cf. in names)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	154 print " -t<g<number>\|s<number>> = minimum number of sequences (g; default = 0) or species (s; default = 0)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	155 print " a gene must have to be retained; both can be specified simultaneously\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	156 print " -h = print this message and quit\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	157 print " -v = verbose output\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	158 exit(0);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	159 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	160 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	161 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	162 print "Don't understand argument: $ARGV[$i]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	163 print "Usage: GenBankStrip.pl -f<filename> [-g<filename>] [-k<number>] [-l<number>] [-o<n\|s>] [-s] [-t] [-h] [-v]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	164 exit(1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	165 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	166 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	167
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	168 die "ERROR: Must supply name of GenBank output file using flag -f.\n" if (not $gbFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	169
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	170 # Load in hardwired gene synonyms
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	171 geneSynonyms();
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	172
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	173 # Get list of target genes (if desired)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	174 if ($geneFile)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	175 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	176 my $userGeneCount = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	177 my %userGenePresent;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	178 setLineBreak($geneFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	179 open (GENE,"<$geneFile") or die "Cannot open file $geneFile containing gene names.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	180 print "Gene(s) to be stripped:\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	181 while (<GENE>)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	182 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	183 chomp;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	184 next unless ($_);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	185 my $gene = $_;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	186 $gene = $synonym{$gene} if (defined $synonym{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	187 $userGene{$gene} = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	188 unless ($userGenePresent{$gene})
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	189 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	190 $userGeneCount++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	191 $userGenePresent{$gene}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	192 print "\t$gene\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	193 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	194 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	195 close GENE;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	196
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	197 # die "ERROR: No genes read in from file $geneFile\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	198 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	199 #THO added -h command to easilly pass a single gene for Galaxy
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	200 if($singleGene)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	201 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	202 my $userGeneCount = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	203 my %userGenePresent;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	204 print "Gene(s) to be stripped:\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	205 my $gene = $singleGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	206 $gene = $synonym{$gene} if (defined $synonym{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	207 print "\t$gene\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	208 $geneFile = $singleGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	209 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	210
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	211 # Print parameter summary
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	212 print "The following parameters have been set by the user:\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	213 print "\tFile containing GenBank sequence data: $gbFile\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	214 print "\tFile containing target genes to be stripped: $geneFile\n" if ($geneFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	215 print "\tUser-defined constraints(s):\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	216 print "\t\tNumber of sequences: $seqThreshold\n" if ($seqThreshold);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	217 print "\t\tNumber of species: $speciesThreshold\n" if ($speciesThreshold);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	218 print "\t\tMinimum sequence length: global - $minLength bp; tRNAs - $minLengthTRNA bp\n" if (defined $minLength);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	219 print "\t\tOnly using species with valid names\n" if ($speciesBlock);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	220 print "\t\tNone\n" if (not $seqThreshold and not $speciesThreshold and not defined $minLength and not $speciesBlock);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	221 print "\tNumber of sequences to keep per species for each gene: ";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	222 if ($keepGene)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	223 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	224 print "$keepGene\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	225 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	226 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	227 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	228 print "all\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	229 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	230 print "\tOutput format(s): fasta";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	231 print ", Se-Al" if ($sealPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	232 print ", nexus" if ($nexusPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	233 print ", phytab" if ($phytabPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	234 print "\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	235
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	236 # Do quick gene count if thresholds indicated; takes longer but 1) saves many disk operations and 2) less memory intensive
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	237 geneCount($gbFile) if (not defined $geneFile and ($seqThreshold or $speciesThreshold)); # Don't bother if user gene list given; will usually be small enough so that benefits won't come into play
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	238
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	239 # Read in GenBank file and strip genes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	240 my $stripZero = time;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	241 print "\nProcessing GenBank file $gbFile ...\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	242 setLineBreak($gbFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	243 open (DATA, "<$gbFile") or die "Cannot open GenBank output file $gbFile\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	244 my @allAccNum;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	245 my ($accNum, %accRead, $duplEntry, $organism, %species, $geneName);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	246 my $speciesFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	247 my (%genePresent, @geneList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	248 my (@startList, @stopList, $complementFlag, $typeStatus, %pseudoStatus, %seqType, $joinLine, @joinSegments, %geneStart, %geneStop, %compStatus, $fullSeq);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	249 my $nameFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	250 my $joinFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	251 my $pseudoFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	252 my $readSeqs = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	253
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	254 while (<DATA>)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	255 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	256 chomp;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	257 my $readLine = $_;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	258 next if ($readLine =~ /^\sLOCUS/ or $readLine =~ /^\sDEFINITION/ or $readLine =~ /^\sVERSION/ or $readLine =~ /^\sKEYWORDS/ or $readLine =~ /^\sSOURCE/ or $readLine =~ /^\sORGANISM/ or $readLine =~ /^\sREFERENCE/ or $readLine =~ /^\sAUTHORS/ or $readLine =~ /^\sTITLE/ or $readLine =~ /^\sJOURNAL/ or $readLine =~ /^\sMEDLINE/ or $readLine =~ /^\sPUBMED/ or $readLine =~ /^\sFEATURES/ or $readLine =~ /^\sCOMMENT/ or $readLine =~ /^\sBASE COUNT/ or $readLine =~ /^\ssource/ or $readLine =~ /^\s\/codon/ or $readLine =~ /^\s\/transl/ or $readLine =~ /^\s\/db_/ or $readLine =~ /^\sCONTIG/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	259
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	260 # Get accession number
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	261 if ($readLine =~ /^\s*ACCESSION\s+(.+)/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	262 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	263 $accNum = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	264 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	265 # Clear variables
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	266 undef @geneList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	267 undef %genePresent;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	268 undef $fullSeq;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	269 undef @startList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	270 undef @stopList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	271 undef %geneStart;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	272 undef %geneStop;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	273 undef %pseudoStatus;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	274 undef $geneName;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	275 $speciesFlag = $nameFlag = $joinFlag = $pseudoFlag = $readSeqs = $duplEntry = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	276
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	277 if (not $accRead{$accNum}) # Check for duplicate entries
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	278 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	279 $accRead{$accNum} = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	280 push @allAccNum, $accNum;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	281 if (scalar(@allAccNum) == int(scalar(@allAccNum)/10000)*10000 and $verbose)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	282 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	283 print "\tSequences read in: ".scalar(@allAccNum)."\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	284 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	285 print "\tAccession number: $accNum\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	286 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	287 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	288 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	289 print "*****NOTE -- DUPLICATE ENTRY Accession $accNum\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	290 $duplEntry = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	291 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	292 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	293
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	294 # Get organism name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	295 if ($readLine =~ /^\s*\/organism=\"(.+)\"/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	296 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	297 $organism = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	298 $organism =~ s/\s/_/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	299 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	300 $species{$accNum} = $organism;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	301 $speciesFlag = 1 if ($organism =~ /sp\./ or $organism =~ /cf\./ or $organism =~ /_X_/i);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	302 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	303 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	304 print "\t\tOrganism: $organism";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	305 print " (blocked)" if ($speciesFlag and $speciesBlock);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	306 print "\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	307 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	308 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	309 next if ($speciesFlag and $speciesBlock); # Entry pertains to invalid species name; skip parsing rest of entry
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	310
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	311
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	312 # Get gene boundaries; process previous set of CDs
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	313 if ($readLine =~ /\<?(\d+)\<?\.\.\>?(\d+)\>?/ or $joinFlag == 1 or $readLine =~ /^\s*ORIGIN/) # ORIGIN will process last set of CDs
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	314 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	315 next if ($readLine =~ /^\s+\/\w+/); # Prevents spurious matches with lines beginning with "/feature"
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	316 $readSeqs = 1 if ($readLine =~ /^\s*ORIGIN/); # Indicates that remaining lines will contain sequence information
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	317
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	318 # Process previous gene; need to do here to account for a posteriori declarations of pseudogene status
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	319 if ($geneName and $nameFlag == 2 and @startList and @stopList) # Process complete gene
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	320 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	321 print "\t\t\t\tParsed name: $geneName (type: $typeStatus)\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	322
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	323 # Clean up gene name and misleading punctuation
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	324 $geneName = geneClean($geneName);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	325 $pseudoStatus{$geneName} = 1 if ($pseudoFlag);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	326
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	327 if (defined @ { $geneStart{$geneName} } and ((defined $pseudoStatus{$geneName} and $pseudoStatus{$geneName} == 1) or $typeStatus =~ /intron/i or $typeStatus =~ /UTR/i)) # Gene has previously stored CDs that might not have been recognized as a pseudogene or non-coding region
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	328 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	329 print "\t\t\t\t\tSubsequent occurrence of now recognized pseudogene or non-coding region; comparing new and stored CDs\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	330 for (my $i = 0; $i < scalar(@startList); $i++) # Check each occurrence in new CDs for matches in stored ones
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	331 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	332 my $newStart = $startList[$i];
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	333 my $newStop = $stopList[$i];
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	334 print "\t\t\t\t\t\tChecking new CDs $newStart to $newStop\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	335 for (my $j = 0; $j < scalar(@ { $geneStart{$geneName} }); $j++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	336 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	337 if ($newStart == $geneStart{$geneName}[$j] and $newStop == $geneStop{$geneName}[$j])
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	338 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	339 print "\t\t\t\t\t\t\tMatch with stored CDs (no. $j); deleted\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	340 splice(@ { $geneStart{$geneName} }, $j, 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	341 splice(@ { $geneStop{$geneName} }, $j, 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	342 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	343 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	344 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	345 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	346 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	347 print "\n\t\t\t\t\tCurrent gene boundaries after pseudogene / non-coding check (type = $seqType{$geneName}):\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	348 if (scalar(@ { $geneStart{$geneName} }) < 1)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	349 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	350 print "\t\t\t\t\t\tnone\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	351 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	352 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	353 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	354 for (my $i = 0; $i < scalar(@ { $geneStart{$geneName} }); $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	355 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	356 print "\t\t\t\t\t\t$geneStart{$geneName}[$i]\t$geneStop{$geneName}[$i]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	357 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	358 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	359 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	360 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	361
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	362 # Only process coding regions of user-desired genes (if applicable), genes with sensible CDs, non-blocked genes, and genes that are not obvious singletons or pseudogenes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	363 unless (($geneFile and not defined $userGene{$geneName}) or
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	364 (defined $geneStatus{$geneName} and $geneStatus{$geneName} eq "rejected") or
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	365 (defined $pseudoStatus{$geneName} and $pseudoStatus{$geneName} == 1) or
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	366 $geneName =~ /hypothetical/i or $geneName =~ /open reading frame/i or $geneName =~ /similar/i or $geneName =~ /homolog/i or $geneName =~ /putative/i or $geneName =~ /unknown/i or $geneName =~ /unnamed/i or $geneName =~ /\d+rik/i or $geneName =~ /possible/i or $geneName =~ /pseudo/i
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	367 or $typeStatus =~ /UTR/i or $typeStatus =~ /intron/i or $typeStatus =~ /misc_feature/i)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	368 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	369 if (not defined $genePresent{$geneName}) # Process first occurrence of gene in entry
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	370 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	371 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	372 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	373 print "\t\t\t\t\tFirst occurrence of $geneName\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	374 for (my $i = 0; $i < scalar(@startList); $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	375 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	376 print "\t\t\t\t\t\t$startList[$i]\t$stopList[$i]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	377 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	378 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	379 $genePresent{$geneName} = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	380 push @geneList, $geneName;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	381 push @ { $geneStart{$geneName} }, $_ foreach (@startList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	382 push @ { $geneStop{$geneName} }, $_ foreach (@stopList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	383
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	384 $seqType{$geneName} = $typeStatus;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	385 $compStatus{$geneName} = 0; # Note whether gene is complemented or not
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	386 $compStatus{$geneName} = 1 if ($complementFlag);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	387 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	388 else # Attempt to add secondary occurrences
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	389 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	390 my $storedSegments = scalar(@ { $geneStop{$geneName} }) - 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	391 my $lastStop = $geneStop{$geneName}[$storedSegments];
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	392 $lastStop = 0 if (not defined $lastStop);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	393 my $newStart = $startList[0];
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	394
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	395 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	396 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	397 print "\t\t\t\t\tSecondary occurrence of $geneName with boundaries\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	398 for (my $i = 0; $i < scalar(@startList); $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	399 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	400 print "\t\t\t\t\t\t$startList[$i]\t$stopList[$i]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	401 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	402 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	403 if ($seqType{$geneName} eq "gene" and $typeStatus ne "gene") # New information probably more precise and better accounts for structure
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	404 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	405 print "\n\t\t\t\t\t\tNew segment more precisely defined by type; replaced\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	406 undef @ { $geneStart{$geneName} };
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	407 undef @ { $geneStop{$geneName} };
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	408 push @ { $geneStart{$geneName} }, $_ foreach (@startList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	409 push @ { $geneStop{$geneName} }, $_ foreach (@stopList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	410 $seqType{$geneName} = $typeStatus;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	411 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	412
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	413 elsif ($newStart > $lastStop) # New segment occurs distal to last stored segment (could also be contiguous); append to boundaries
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	414 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	415 print "\n\t\t\t\t\t\tContiguous with or occurs after last stored segment; appended\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	416 push @ { $geneStart{$geneName} }, $_ foreach (@startList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	417 push @ { $geneStop{$geneName} }, $_ foreach (@stopList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	418 $seqType{$geneName} = "composite";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	419 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	420 elsif (scalar(@ { $geneStart{$geneName} }) == 1 and scalar(@startList) > 1) # Replace single stored segment with new segments derived from join statement; probably better accounts for intron/exon structure
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	421 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	422 print "\n\t\t\t\t\t\tNew segments subdivide single stored segment; replaced\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	423 undef @ { $geneStart{$geneName} };
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	424 undef @ { $geneStop{$geneName} };
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	425 push @ { $geneStart{$geneName} }, $_ foreach (@startList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	426 push @ { $geneStop{$geneName} }, $_ foreach (@stopList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	427 $seqType{$geneName} = "composite";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	428 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	429 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	430 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	431 print "\n\t\t\t\t\t\tNew segment overlaps stored segments; rejected\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	432 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	433 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	434 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	435 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	436 print "\n\t\t\t\t\tCurrent gene boundaries after CDS processing (type = $seqType{$geneName}):\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	437 if (scalar(@ { $geneStart{$geneName} }) < 1)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	438 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	439 print "\t\t\t\t\t\tnone\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	440 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	441 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	442 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	443 for (my $i = 0; $i < scalar(@ { $geneStart{$geneName} }); $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	444 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	445 print "\t\t\t\t\t\t$geneStart{$geneName}[$i]\t$geneStop{$geneName}[$i]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	446 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	447 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	448 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	449 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	450
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	451 # Clear entries for gene
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	452 $nameFlag = $complementFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	453 undef @startList; # Prevents double listing of genes if CDS use both /gene and /product tags
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	454 undef @stopList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	455 undef $typeStatus;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	456 $pseudoStatus{$geneName} = 0; # Clear pseudogene status for previous set of CDs
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	457 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	458 next if ($readLine =~ /^\s*ORIGIN/); # No more CDs to worry about
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	459
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	460 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	461
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	462 # Reset gene name and information for current set of CDs
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	463 undef $geneName;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	464 $nameFlag = $pseudoFlag = $complementFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	465 $complementFlag = 1 if ($readLine =~ /complement/); #DISABLED BY THO BECAUSE ONLY COMPLEMENTS AND DOESN'T REVERSE
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	466
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	467 # Get new CD information
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	468 if ($readLine =~ /^\s+\S+\s+join\(/i or $readLine =~ /^\s+\S+\s+order\(/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	469 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	470 $joinFlag = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	471 undef @startList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	472 undef @stopList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	473 undef $joinLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	474 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	475 if ($readLine =~ /^\s+(\S+)\s+/i)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	476 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	477 $typeStatus = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	478 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	479
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	480 if ($joinFlag)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	481 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	482 $joinLine .= $readLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	483 if ($readLine =~ /\)$/) # Have accounted for multiline join statements; process
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	484 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	485 $joinFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	486 $complementFlag = 1 if ($joinLine =~ /complement/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	487
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	488 # Clean up join statement
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	489 $joinLine =~ s/complement\(//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	490 $joinLine =~ s/>//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	491 $joinLine =~ s/<//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	492 $joinLine =~ s/^\s+\S+\s+join\(//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	493 $joinLine =~ s/^\s+\S+\s+order\(//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	494 $joinLine =~ s/\)+$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	495 $joinLine =~ s/\s+//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	496 print "\t\tJoin boundaries: $joinLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	497 if ($joinLine =~ /[A-Z]+\d+/i) # Avoid join commands referring to separate accessions
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	498 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	499 undef $joinLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	500 print "\t\t\tERROR: Joining accessions; skipping\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	501 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	502 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	503
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	504 # Derive pieces of join statement
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	505 @joinSegments = split(/,/, $joinLine);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	506 foreach my $segment (@joinSegments)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	507 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	508 my ($start, $stop) = split(/\.\./, $segment);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	509 # Check safeties
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	510 next unless ($start and $stop);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	511 next if ($start =~ /\D/ or $stop =~ /\D/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	512 next if ($start > $stop);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	513 push @startList, $start;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	514 push @stopList, $stop;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	515 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	516
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	517 undef $joinLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	518 $typeStatus = "join";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	519 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	520 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	521 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	522 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	523 my ($start, $stop);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	524 if ($readLine =~ /\<?(\d+)\<?\.\.\>?(\d+)\>?/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	525 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	526 $start = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	527 $stop = $2;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	528 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	529 next unless ($start and $stop);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	530 next if ($start > $stop);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	531 undef @startList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	532 push @startList, $start;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	533 undef @stopList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	534 push @stopList, $stop;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	535 print "\t\tParsed boundaries: $1\t$2\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	536 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	537 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	538
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	539
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	540 # Check for pseudogene status
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	541 if ($readLine =~ /\s+\/pseudo/ or $readLine =~ /pseudogene/ or $readLine =~ /putative/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	542 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	543 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	544 $pseudoStatus{$geneName} = 1 if (defined $geneName);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	545 $pseudoFlag = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	546 print "\t\t\t\tWARNING: suspected pseudogene or of uncertain (\"putative\") status\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	547 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	548
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	549 if ($readLine =~ /codon recognized/ and defined $geneName and $geneName =~ /trna/i)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	550 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	551 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	552 $pseudoStatus{$geneName} = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	553 $pseudoFlag = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	554 print "\t\t\t\tWARNING: tRNA for an alternative codon; ignored\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	555 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	556
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	557 # Get gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	558 if ($readLine =~ /^\s*(\/gene)\|(\/product)=\"(.+)/ and $nameFlag == 0) # Get start of gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	559 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	560 $geneName = $1 if ($readLine =~ /=\"(.+)/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	561 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	562
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	563 # Check whether name wraps onto a new line
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	564 if (substr($readLine, -1) ne "\"")
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	565 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	566 $nameFlag = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	567 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	568 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	569 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	570 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	571 $nameFlag = 2;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	572 $geneName =~ s/\"$// if ($geneName =~ /\"$/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	573 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	574 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	575
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	576 if ($nameFlag == 1) # Get continuation / end of gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	577 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	578 print "$readLine\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	579 $geneName .= $readLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	580 $nameFlag = 2 if ($readLine =~ /\"$/) # Gene name is complete
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	581 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	582
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	583 # Read in sequence information and append
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	584 if ($readLine =~ /^\s\d1\s\w+/ and @geneList and $readSeqs)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	585 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	586 my $seqFrag = $readLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	587 $seqFrag =~ s/\s+//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	588 $seqFrag =~ s/\d+//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	589 $fullSeq .= uc($seqFrag);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	590 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	591
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	592 # End of entry; process
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	593 if ($readLine =~ /\/\//)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	594 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	595 next if ($duplEntry);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	596 next unless (@geneList and defined $fullSeq);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	597 foreach my $gene (@geneList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	598 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	599 # Safeties; shouldn't come into play
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	600 next if ($geneFile and not defined $userGene{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	601 next if (defined $geneStatus{$gene} and $geneStatus{$gene} eq "rejected");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	602 next if (defined $pseudoStatus{$gene} and $pseudoStatus{$gene} == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	603
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	604 print "\n\t\tProcessing gene $gene\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	605
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	606 # Strip gene out of full sequence and format as fasta
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	607 next unless (@ { $geneStart{$gene} } and @ { $geneStop{$gene} } );
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	608
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	609 my $geneSeq;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	610 for (my $i = 0; $i < scalar(@ { $geneStart{$gene} }); $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	611 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	612 print "\t\t\tGetting sequence between positions $geneStart{$gene}[$i] and $geneStop{$gene}[$i]\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	613 next if ($geneStart{$gene}[$i] > length($fullSeq) or $geneStop{$gene}[$i] > length($fullSeq));
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	614 my $geneSegment = substr($fullSeq,$geneStart{$gene}[$i]-1,$geneStop{$gene}[$i]-$geneStart{$gene}[$i]+1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	615 print "\t\t\t\t$geneSegment\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	616 $geneSeq .= $geneSegment;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	617 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	618
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	619 next unless ($geneSeq);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	620 $geneSeq = complement($geneSeq) if ($compStatus{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	621 $maxLength{$gene} = length($geneSeq) if (not defined $maxLength{$gene} or length($geneSeq) > $maxLength{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	622
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	623 # Check if sequence matches length threshold (if appropriate)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	624 if (defined $minLength)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	625 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	626 if ($gene =~ /^trna-\w\w\w/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	627 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	628 if (length($geneSeq) < $minLengthTRNA)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	629 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	630 printf "\t\t\tRejected: length (%s bp) does not meet tRNA length threshold ($minLengthTRNA bp)\n", length($geneSeq) if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	631 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	632 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	633 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	634 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	635 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	636 if (length($geneSeq) < $minLength)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	637 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	638 printf "\t\t\tRejected: length (%s bp) does not meet global threshold ($minLength bp)\n", length($geneSeq) if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	639 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	640 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	641 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	642 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	643 if ($gene =~ /trna-\w\w\w/ and length($geneSeq) > 100)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	644 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	645 printf "\t\t\tRejected: length of tRNA too long (%s bp); indicates parsing failure\n", length($geneSeq) if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	646 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	647 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	648
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	649 my $breakPoint = 79;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	650 until ($breakPoint > length($geneSeq))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	651 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	652 my $replaceString = "\n" . substr($geneSeq, $breakPoint, 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	653 substr($geneSeq, $breakPoint, 1) = $replaceString;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	654 $breakPoint += 80;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	655 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	656
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	657 # Append sequence to file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	658 $geneStatus{$gene} = "stripped";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	659 $sequenceCount{$gene}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	660 $sequenceCount{$species{$accNum}}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	661 $sequenceCount{"all"}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	662 my $fastaFile = $gene."_gbs.fasta.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	663
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	664 my $IOicon = ">";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	665 $IOicon .= ">" if ($sequenceCount{$gene} > 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	666
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	667 unless ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	668
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	669
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	670 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	671 open (GENE, $IOicon, $fastaFile) or die "Cannot open file $fastaFile to write sequence to.";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	672 print GENE ">$species{$accNum} $accNum\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	673 print GENE "$geneSeq\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	674 close GENE;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	675 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	676
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	677 # Update various statistical counters
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	678 unless ($globalGenePresent{$gene}) # Gene counter
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	679 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	680 push @globalGeneList, $gene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	681 $globalGenePresent{$gene} = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	682 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	683
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	684 unless ($speciesPresent{$species{$accNum}}) # Species counter
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	685 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	686 push @speciesList, $species{$accNum};
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	687 $speciesPresent{$species{$accNum}} = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	688 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	689
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	690 $speciesGenePresent{$gene}{$species{$accNum}}++; # Species-gene counter
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	691 $speciesCount{$gene}++ if ($speciesGenePresent{$gene}{$species{$accNum}} == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	692
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	693 # Print out summary stats for gene
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	694 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	695 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	696 print "\t\t\tGene: $gene\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	697 printf "\t\t\t\tGene boundaries (%s bp)", length($geneSeq) - ($geneSeq =~ tr/\n//);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	698 print " (complemented)" if ($compStatus{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	699 print ":\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	700 for (my $i = 0; $i < scalar(@ { $geneStart{$gene} }); $i++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	701 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	702 print "\t\t\t\t\t$geneStart{$gene}[$i]\t$geneStop{$gene}[$i]\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	703 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	704 print "\t\t\t\t\t$geneSeq\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	705 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	706 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	707 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	708 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	709 close DATA;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	710
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	711 # Print out summary stats
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	712 print "\n\tSummary stats:\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	713 printf "\t\tTotal number of accessions processed: %s\n", scalar(@allAccNum);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	714 printf "\t\tTotal number of unique species: %s\n", scalar(@speciesList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	715 printf "\t\tTotal number of unique sequences: %s\n", $sequenceCount{"all"};
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	716 printf "\t\tTotal number of unique genes: %s\n", scalar(@globalGeneList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	717
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	718 my $stripTime = time - $stripZero;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	719 print "\n\t\tTime taken to process file: $stripTime seconds\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	720
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	721 if (not @globalGeneList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	722 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	723 print "\nNOTE: No genes stripped from file $gbFile; exiting\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	724 exit(0);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	725 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	726
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	727 # Reprocess results to 1) recheck whether stripped genes actually fall below user-set thresholds (geneCount procedure is a rough maximum), 2) pare down to required number of sequences per species, and 3) produce appropriate output files
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	728 print "\nPostprocessing results ...\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	729 my $postZero = time;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	730
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	731 @globalGeneList = sort(@globalGeneList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	732 $stripCount = scalar(@globalGeneList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	733
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	734 foreach my $gene (@globalGeneList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	735 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	736 next if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	737 print "\tProcessing gene $gene ($sequenceCount{$gene} sequences for $speciesCount{$gene} species)\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	738 if ($sequenceCount{$gene} < $seqThreshold or $speciesCount{$gene} < $speciesThreshold) # Gene does not meet threshold requirements; reject
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	739 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	740 $geneStatus{$gene} = "rejected";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	741 push @rejectList, $gene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	742 $stripCount--;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	743 print "\t\tRejected; does not meet threshold requirements\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	744
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	745 # Remove associated file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	746 my $rmString = $gene."_gbs.fasta.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	747 $rmString =~ s/\s/\\ /g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	748 $rmString =~ s/\(/\\(/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	749 $rmString =~ s/\)/\\)/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	750 $rmString =~ s/\'/\\'/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	751 system ("rm $rmString");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	752 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	753 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	754 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	755 my (%infLength, %lengthList, %criticalLength);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	756 undef @speciesList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	757 $geneStatus{$gene} = "stripped";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	758 if ($verbose)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	759 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	760 print "\t\tStripped";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	761 print "; meets threshold requirements" if ($seqThreshold or $speciesThreshold);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	762 print "\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	763 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	764
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	765 # Reload sequences from disk
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	766 my $inputFile = $gene."_gbs.fasta.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	767 setLineBreak($inputFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	768 open (FASTA, "<$inputFile") or die "Cannot open file $inputFile to read data for gene $gene\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	769 my (@accList, %speciesName, %fastaSeq, $species, $fastaAcc);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	770
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	771 undef %speciesPresent;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	772 while (<FASTA>)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	773 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	774 chomp;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	775 if ($_ =~ "^>")
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	776 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	777 ($species, $fastaAcc) = split;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	778 $species =~ s/^>//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	779 $fastaAcc =~ s/^\(//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	780 $fastaAcc =~ s/\)$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	781
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	782
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	783 push @accList, $fastaAcc;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	784 $speciesName{$fastaAcc} = $species;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	785 $speciesPresent{$species}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	786 push @speciesList, $species if ($speciesPresent{$species} == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	787 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	788 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	789 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	790 $fastaSeq{$fastaAcc} .= $_;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	791 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	792 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	793 close FASTA;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	794 @accList = sort { $speciesName{$a} cmp $speciesName{$b} } keys %speciesName;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	795
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	796 # Pare down to desired number of sequences as needed
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	797 if ($keepGene)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	798 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	799 undef %lengthList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	800 print "\t\tParing down to $keepGene best lengths for each of $speciesCount{$gene} species ...\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	801 # Get informative length for each sequence
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	802 foreach my $entry (@accList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	803 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	804 $infLength{$entry} = ($fastaSeq{$entry} =~ tr/ACGT//);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	805 push @ { $lengthList{$speciesName{$entry}} }, $infLength{$entry};
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	806 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	807
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	808 # Determine critical length and correct sequence count for number of deleted species
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	809 foreach my $species (@speciesList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	810 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	811 print "\t\t\tSpecies: $species ($speciesPresent{$species} sequences)\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	812 next if ($speciesPresent{$species} <= $keepGene); # Only process species for which gene has been sampled more than threshold
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	813 @ { $lengthList{$species} } = sort {$b <=> $a} @ { $lengthList{$species} }; # Sort in descending order and get critical length
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	814 $criticalLength{$species} = $lengthList{$species}[$keepGene-1];
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	815
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	816 # Correct sequence count
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	817 my $i = $keepGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	818 $i++ until ($i >= $speciesPresent{$species} or $lengthList{$species}[$i] < $criticalLength{$species});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	819 $sequenceCount{$gene} -= (scalar(@ { $lengthList{$species} }) - $i);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	820
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	821 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	822 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	823 print "\t\t\t\tCritical length: $criticalLength{$species}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	824 printf "\t\t\t\tNumber of sequences deleted: %s\n", scalar(@ { $lengthList{$species} }) - $i;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	825 print "\t\t\t\tTotal number of sequences for gene remaining: $sequenceCount{$gene}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	826 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	827 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	828
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	829 print "\t\t\tFinal count: $sequenceCount{$gene} sequences\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	830
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	831 # Recheck if paring has dropped sequence below sequence threshold
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	832 if ($sequenceCount{$gene} < $seqThreshold)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	833 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	834 print "\t\t\tNOTE: Gene ($sequenceCount{$gene} sequences) now falls below threshold of $seqThreshold sequences; no file created\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	835 $geneStatus{$gene} = "rejected";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	836 push @rejectList, $gene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	837 $stripCount--;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	838
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	839 # Remove associated file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	840 my $rmString = $gene."_gbs.fasta.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	841 $rmString =~ s/\s/\\ /g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	842 $rmString =~ s/\(/\\(/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	843 $rmString =~ s/\)/\\)/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	844 $rmString =~ s/\'/\\'/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	845 system ("rm $rmString");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	846
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	847 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	848 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	849 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	850
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	851 # Produce appropriate output files
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	852 print "\t\tSaving output ...\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	853 my %printCount;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	854
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	855 # Open files as needed
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	856 my $sealFile = $gene."_gbs.seal";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	857 open (SEAL, ">$sealFile") or die "Cannot open Se-Al formatted file $sealFile for writing\n" if ($sealPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	858 my $nexusFile = $gene."_gbs.nex";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	859 open (NEX, ">$nexusFile") or die "Cannot open nexus-formatted file $nexusFile for writing\n" if ($nexusPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	860 # my $phytabFile = $gene."_gbs.phytab";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	861 # open (PHYTAB, ">$phytabFile") or die "Cannot open nexus-formatted file $phytabFile for writing\n" if ($phytabPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	862 #Currently assumes that only 1 file will be written for ease of Galaxy implementation
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	863 my $phytabFile = "osiris_gbs.phytab";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	864 open (PHYTAB, ">$phytabFile") or die "Cannot open nexus-formatted file $phytabFile for writing\n" if ($phytabPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	865 my $fastaFile = $gene."_gbs.fasta.txt";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	866 $fastaFile = $gene."_gbs.new.fasta.txt" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	867 open (FASTA, ">$fastaFile") or die "Cannot open fasta-formatted file $fastaFile for writing\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	868
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	869 # Print headers
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	870 if ($sealPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	871 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	872 print SEAL "Database={\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	873 print SEAL "\tID='MLst';\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	874 print SEAL "\tOwner=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	875 print SEAL "\tName=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	876 print SEAL "\tDescription=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	877 print SEAL "\tFlags=0;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	878 print SEAL "\tCount=2;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	879 print SEAL "\t{\n\t\t{\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	880
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	881 print SEAL "\t\t\tID='PAli';\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	882 print SEAL "\t\t\tOwner=1;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	883 printf SEAL "\t\t\tName=\"%sgbs\";\n", $gene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	884 print SEAL "\t\t\tDescription=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	885 print SEAL "\t\t\tFlags=0;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	886 print SEAL "\t\t\tNumSites=$maxLength{$gene};\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	887 print SEAL "\t\t\tType=\"Nucleotide\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	888 print SEAL "\t\t\tFeatures=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	889 print SEAL "\t\t\tColourMode=1;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	890 print SEAL "\t\t\tLabelMode=0;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	891 print SEAL "\t\t\ttriplets=false;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	892 print SEAL "\t\t\tinverse=true;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	893 printf SEAL "\t\t\tCount=%s;\n", $sequenceCount{$gene};
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	894 print SEAL "\t\t\t{\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	895 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	896
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	897 if ($nexusPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	898 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	899 print NEX "#NEXUS\n\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	900 print NEX "Begin data;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	901 printf NEX "\tDimensions ntax=%s nchar=%s;\n", $sequenceCount{$gene}, $maxLength{$gene};
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	902 print NEX "\tFormat datatype=nucleotide gap=-;\n\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	903 print NEX "\tmatrix\n\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	904 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	905
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	906 # Print sequence data
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	907 my $i = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	908 foreach my $entry (@accList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	909 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	910 next if (defined $criticalLength{$speciesName{$entry}} and $infLength{$entry} < $criticalLength{$speciesName{$entry}});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	911
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	912 $printCount{$speciesName{$entry}}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	913 $speciesName{$entry} .= "_$printCount{$speciesName{$entry}}" if ($printCount{$speciesName{$entry}} > 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	914
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	915 if ($sealPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	916 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	917 $i++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	918 print SEAL "\t\t\t\t{\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	919 print SEAL "\t\t\t\t\tID='PSeq';\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	920 print SEAL "\t\t\t\t\tOwner=2;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	921 print SEAL "\t\t\t\t\tName=\"$speciesName{$entry}\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	922 print SEAL "\t\t\t\t\tDescription=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	923 print SEAL "\t\t\t\t\tFlags=0;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	924 print SEAL "\t\t\t\t\tAccession=\"$entry\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	925 print SEAL "\t\t\t\t\tType=\"DNA\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	926 printf SEAL "\t\t\t\t\tLength=%s;\n", length($fastaSeq{$entry});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	927 print SEAL "\t\t\t\t\tSequence=\"$fastaSeq{$entry}\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	928 print SEAL "\t\t\t\t\tGeneticCode=1;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	929 print SEAL "\t\t\t\t\tCodeTable=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	930 print SEAL "\t\t\t\t\tFrame=1;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	931 print SEAL "\t\t\t\t\tFeatures=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	932 print SEAL "\t\t\t\t\tParent=null;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	933 print SEAL "\t\t\t\t\tComplemented=false;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	934 print SEAL "\t\t\t\t\tReversed=false;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	935 print SEAL "\t\t\t\t}";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	936 print SEAL "," unless ($i == $sequenceCount{$gene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	937 print SEAL "\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	938 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	939
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	940 if ($nexusPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	941 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	942 my $gaps = "-" x ($maxLength{$gene} - length($fastaSeq{$entry}));
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	943 print NEX "$speciesName{$entry}\t$fastaSeq{$entry}$gaps\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	944 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	945 if($phytabPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	946 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	947 print PHYTAB "$speciesName{$entry} \t$gene \t $entry \t $fastaSeq{$entry} \n"
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	948 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	949
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	950 my $breakPoint = 79;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	951 until ($breakPoint > length($fastaSeq{$entry}))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	952 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	953 my $replaceString = "\n" . substr($fastaSeq{$entry}, $breakPoint, 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	954 substr($fastaSeq{$entry}, $breakPoint, 1) = $replaceString;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	955 $breakPoint += 80;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	956 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	957 if ($accessionFirst) #ADDED BY THO
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	958 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	959 print FASTA ">$entry", "\___", "$speciesName{$entry}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	960 print FASTA "$fastaSeq{$entry}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	961 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	962 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	963 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	964 print FASTA ">$speciesName{$entry} ($entry)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	965 print FASTA "$fastaSeq{$entry}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	966 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	967 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	968 # Print footers
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	969 if ($sealPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	970 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	971 print SEAL "\t\t\t};\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	972 print SEAL "\t\t},\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	973 print SEAL "\t\t{\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	974 print SEAL "\t\t\tID='MCoL';\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	975 print SEAL "\t\t\tOwner=1;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	976 print SEAL "\t\t\tName=\"Genetic Codes\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	977 print SEAL "\t\t\tDescription=\"Custom Genetic Codes\";\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	978 print SEAL "\t\t\tFlags=0;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	979 print SEAL "\t\t\tCount=0;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	980 print SEAL "\t\t}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	981 print SEAL "\t};\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	982 print SEAL "};\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	983 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	984 if ($nexusPrint)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	985 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	986 print NEX "\t;\nend;\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	987 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	988
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	989 # Close files as appropriate
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	990 close SEAL if ($sealPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	991 close NEX if ($nexusPrint);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	992 close FASTA if ($keepGene);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	993 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	994 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	995
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	996 my $postTime = time - $postZero;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	997 print "\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	998 print "\tTime taken: $postTime seconds\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	999
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1000 # Print out final summary stats / files
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1001 print "\nFinal summary statistics\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1002
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1003 # Print file of stripped genes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1004 open (STRIP, ">$stripFile") or die "Cannot write summary of stripped genes to $stripFile.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1005 print STRIP "The following $stripCount genes were stripped successfully from file $gbFile\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1006 print STRIP "\nGene\tNo. of sequences\tNo. of species\n\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1007
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1008 foreach my $gene (@globalGeneList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1009 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1010 print STRIP "$gene\t$sequenceCount{$gene}\t$speciesCount{$gene}\n" unless ($geneStatus{$gene} eq "rejected");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1011 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1012 close STRIP;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1013
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1014 print "\tSummaries of $stripCount genes that were stripped from $gbFile have been written to $stripFile\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1015
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1016 # Print file of rejected genes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1017 if (@rejectList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1018 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1019 @rejectList = sort @rejectList;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1020 open (REJECT, ">$rejectFile") or die "Cannot write summary of rejected genes to $rejectFile.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1021 printf REJECT "The following %s genes do not meet user-defined threshold(s) of", scalar(@rejectList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1022 print REJECT " $seqThreshold sequences" if ($seqThreshold);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1023 print REJECT " or" if ($seqThreshold and $speciesThreshold);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1024 print REJECT " $speciesThreshold species" if ($speciesThreshold);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1025 print REJECT "; an asterisk indicates that values given are rough maximums\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1026 print REJECT "\nGene\tNo. of sequences\tNo. of species\n\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1027
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1028 foreach my $gene (@rejectList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1029 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1030 print REJECT "$gene\t";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1031 if (defined $sequenceCount{$gene}) # Need to do alternative versions depending on whether gene was properly counted or not
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1032 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1033 print REJECT "$sequenceCount{$gene}\t";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1034 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1035 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1036 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1037 print REJECT "$quickSequenceCount{$gene} \*\t";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1038 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1039 if (defined $speciesCount{$gene})
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1040 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1041 print REJECT "$speciesCount{$gene}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1042 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1043 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1044 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1045 print REJECT "$quickSpeciesCount{$gene} \*\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1046 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1047 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1048 close REJECT;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1049
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1050 printf "\tSummaries of %s genes that did not meet threshold(s) have been written to $rejectFile\n", scalar(@rejectList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1051 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1052
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1053 if ($debug)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1054 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1055 print "\n\tSummary stats for each processed gene:\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1056 foreach my $gene (@globalGeneList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1057 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1058 print "\t\tGene: $gene ($geneStatus{$gene})\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1059 print "\t\t\tTotal number of sequences: $sequenceCount{$gene}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1060 print "\t\t\tTotal number of unique species: $speciesCount{$gene}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1061 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1062 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1063
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1064 exit(0);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1065
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1066 # Subroutines used in the script
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1067
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1068 sub setLineBreak # Check line breaks of input files and set input record separator accordingly
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1069 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1070 my $gbFile = shift;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1071 $/ ="\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1072 open (IN, "<$gbFile") or die "Cannot open $gbFile to check form of line breaks.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1073 while (<IN>)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1074 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1075 if ($_ =~ /\r\n/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1076 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1077 print "\tDOS line breaks detected ...\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1078 $/ ="\r\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1079 last;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1080 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1081 elsif ($_ =~ /\r/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1082 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1083 print "\tMac line breaks detected ...\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1084 $/ ="\r";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1085 last;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1086 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1087 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1088 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1089 print "\tUnix line breaks detected ...\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1090 $/ ="\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1091 last;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1092 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1093 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1094 close IN;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1095 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1096
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1097 sub complement # Outputs complementary sequence to one provided
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1098 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1099 my $tempSeq = shift;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1100
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1101 my $compSeq;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1102 for (my $nt = 0; $nt < length($tempSeq); $nt++)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1103 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1104 if (not defined $complement{substr($tempSeq, $nt, 1)})
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1105 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1106 $compSeq .= "?";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1107 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1108 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1109 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1110 $compSeq .= $complement{substr($tempSeq, $nt, 1)};
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1111 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1112 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1113 my $revCompSeq = reverse($compSeq);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1114 return $revCompSeq;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1115 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1116
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1117 sub geneSynonyms # Define gene synonyms; originally compiled by Robin Beck
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1118 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1119 #Opsin gene added by THO
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1120 $synonym{$_} = "opsin" foreach ("opsin", "anceropsin", "blop", "blue-sensitive_opsin", "blue-sensitive_opsin_precursor", "blue-sensitive_rhodopsin", "blue-sensitive_visual_pigment", "blue_opsin", "bluerh", "boceropsin", "buvops", "compound_eye_opsin_bcrh1", "compound_eye_opsin_bcrh2", "lateral_eye_opsin", "locus_opsin_1", "locust_opsin_2", "long-wavelenght_opsin", "long-wavelength_like_opsin", "long-wavelength_opsin", "long-wavelength_rhodopsin", "long-wavelength_sensitive_opsin_1", "long-wavelength_sensitive_opsin_2", "long_wave_opsin", "long_wavelength-sensitive_opsin", "long_wavelength-sensitive_rhodopsin", "long_wavelength-sensitive_visual_pigment", "long_wavelength_opsin", "long_wavelength_sensitive_opsin_1", "long_wavelength_sensitive_opsin_2", "lop2", "lw_opsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1121 "ocellar_opsin", "opsin", "opsin_2", "opsin_bcrh1", "opsin_bcrh2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1122 "opsin_rh1", "opsin_rh3", "opsin_rh4", "piceropsin", "pteropsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1123 "rh1_opsin", "rh2_opsin", "rh3_opsin", "rh4_opsin", "rh6_rhodopsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1124 "rhodopsin", "rhodopsin_1", "rhodopsin_2_cg16740-pa", "rhodopsin_3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1125 "rhodopsin_3_cg10888-pa", "rhodopsin_4", "rhodopsin_4_cg9668-pa",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1126 "rhodopsin_5", "rhodopsin_5_cg5279-pa", "rhodopsin_6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1127 "rhodopsin_6_cg5192-pb", "rhodopsin_7_cg5638-pa",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1128 "rhodopsin_long-wavelength", "short_wavelength-sensitive_opsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1129 "ultraviolet-sensitive_opsin", "ultraviolet-sensitive_rhodopsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1130 "ultraviolet-sensitive_visual_pigment", "ultraviolet_sensitive_opsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1131 "uv-sensitive_opsin", "uv-wavelength_like_opsin", "uv_opsin", "uvop",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1132 "uvrh", "uvrh1", "amblop", "amuvop",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1133 "lwrh", "lwrh1", "lwrh2", "lw", "lw-rh", "lw_rh", "ninae", "ninae-pa",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1134 "op", "ops", "ops1", "opsin_1", "opsin_3", "rh", "rh1", "rh2", "rh3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1135 "rh2-pa", "rh3-pa", "rh4", "rh4-pa", "rh5", "rh6", "rh7", "rho",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1136 "visual_pigment", "long-wavelength_rodopsin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1137 "long_wavelength_rhodopsin", "short-wavelength_rhodopsin");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1138
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1139 #Chloroplast genes added by THO***
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1140 $synonym{$_} = "rbcl" foreach ("rbcl", "large_subunit_of_riblose-15-bisphosphate_carboxylase-oxygenase","larger_subunit_of_rubisco",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1141 "rubisco_large_subunit", "ribulose_15-bisphosphate_carboxylase_large_subunit", "ribulosebiphosphate_carboxylase_large_subunit");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1142 $synonym{$_} = "matk" foreach ("matk", "maturase_k", "maturase");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1143
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1144 # Mitochondrial genes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1145 $synonym{$_} = "mtatp6" foreach ("mtatp6", "atp synthase 6", "atp synthase a chain", "atp6", "mt-atp6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1146 "atpase subunit 6", "atpase6", "atpase 6", "atpase6 protein", "atp6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1147 "atp synthase f0 subunit 6", "atp synthase a chain protein 6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1148 "atp synthase subunit 6", "atpase6", "atp6", "atp sythase subunit 6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1149 "atpase subunit 6, atpase6", "f0-atpase subunit 6", "f1atpase subunit 6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1150 "f1-atpase subunit 6", "atpase 6", "atpase 6 gene", "atpase subunit 6 (atp6)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1151 "atpase subunit 6 (atpase6)");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1152 $synonym{$_} = "mtatp8" foreach ("mtatp8", "atp synthase 8", "atp synthase protein 8", "atpase subunit 8",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1153 "a6l", "atp8", "mt-atp8", "atpase subunit 8", "atpase8", "atpase 8",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1154 "atpase8 protein", "atp8", "atp synthase f0 subunit 8", "atp synthase protein 8",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1155 "atpase-8", "atp synthase subunit 8", "atpase8", "atp8", "atp sythase subunit 8",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1156 "atpase subunit8", "f0-atpase subunit 8", "f1 atpase subunit 8", "f1-atpase subunit 8",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1157 "protein a6l", "atpase 8", "atpase subunit 8 (atp8)", "atpase subunit 8 (atpase8)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1158 "atpase 8 gene", "atpase subunit 8 (atp8)", "atpase subunit 8 (atpase8)");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1159 $synonym{$_} = "mtco1" foreach ("mtco1", "cytochrome c oxidase i", "coi", "mt-co1", "co i", "ccoi", "cox 1", "cox i",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1160 "coi", "cytochrome c oxidase subunit i", "cytochrome oxidase subunit i", "cox1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1161 "cytochrome c oxidase subunit 1", "cytochrome oxidase subunit 1", "co1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1162 "cytochrome c oxidase polypeptide i", "cytochrome oxidase i", "cox-1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1163 "cytochrome oxidase c subunit 1", "cox1", "co i", "co1", "cox 1", "coxi",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1164 "cytochrome c oxidase polypeptide i", "cytochrome c-oxidase subunit 1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1165 "cytochrome oxidase c subunit i", "cytochrome-c oxidase i", "cytochrome c1 mrna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1166 "cytochrome c1", "cytochrome c oxidase subunit 1 (coxi)", "cytochrome c oxidase chain i",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1167 "cytochrome c1 (aa 1-241)", "cytochrome c oxidase polypeptide 1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1168 "cytochrome c oxidase subunit 1 (coi)", "cytochrome c-1");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1169 $synonym{$_} = "mtco2" foreach ("mtco2", "cytochrome c oxidase ii", "cytochrome c oxidase polypeptide ii", "coii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1170 "mt-co2", "coii", "cytochrome c oxidase subunit ii", "cytochrome oxidase subunit ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1171 "cytochrome oxidase subunit 2", "cytochrome c oxidase subunit 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1172 "cytochrome c oxidase ii", "cox2", "co2", "cytochrome c oxidase polypeptide ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1173 "cytochrome oxidase ii", "cox2", "cytochrome oxidase c subunit 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1174 "cytochrome-c oxidase", "co ii", "co2", "cox 2", "cytochrome c oxidase polypeptide ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1175 "cytochrome c-oxidase subunit 2", "cytochrome oxidase c subunit ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1176 "cytochrome oxidase subunit2", "cytochrome-c oxidase ii", "cytochrome c oxidase subunit 2 (coxii)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1177 "cytochrome c oxidase subunit 2 (coii)", "cytochrome c oxidase chain ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1178 "cytochrome c oxidase polypeptide 2", "cytochrome c oxidase ii subunit",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1179 "cytochrome c oxidase ii mrna");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1180 $synonym{$_} = "mtco3" foreach ("mtco3", "cytochrome c oxidase iii", "coiii", "mt-co3", "co iii", "ccoiii", "cox3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1181 "cox iii", "cytochrome oxidase subunit iii", "coiii", "cox iii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1182 "cytochrome c oxidase subunit iii", "cytochrome oxidase subunit 3", "cox3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1183 "cytochrome c oxidase subunit 3", "co3", "cytochrome c oxidase polypeptide iii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1184 "cox3", "cytochrome oxidase c subunit 3", "cytochrome oxidase iii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1185 "cytochrome c oxidase polypeptide iii", "co iii", "coiii", "coiii protein", "coxiii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1186 "cytochrome c oxidase subunit iii, coiii", "cytochrome c-oxidase subunit 3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1187 "cytochrome c-oxidase subunit three", "cytochrome oxidase c subunit iii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1188 "cytochrome-c oxidase iii", "cytochrme c oxidase iii", "cytochrome c oxidase subunit 3 (coiii)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1189 "cytochrome oxidase subunit iii type 2");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1190 $synonym{$_} = "mtnd1" foreach ("mtnd1", "nd1", "nadh dehydrogenase 1", "nadh dehydrogenase, subunit 1 (complex i)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1191 "mt-nd1", "nadh-ubiquinone oxidoreductase chain 1", "nd1", "nadh1", "nd1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1192 "nadh-ubiquinone oxidoreductase chain 1", "nadh-ubiquinone oxidoreductase subunit 1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1193 "nadh dehydrogenase i", "nadh dehydrogenase 1", "nadh subunit 1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1194 "nadh dehydrogenase subnuit 1", "nadh1", "nadh1 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1195 "nadh-ubiquinone oxidoreductase subunit i", "nadh dehydrogenase subunit 1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1196 "nadh dehydrogenase subunit 1 (nd1)", "nadh dehydrogenase subunit 1 type 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1197 "nadh dehydrogenase subunit 1 type 1", "nadh dehydrogenase subunit i", "nadh dehydrogenase i",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1198 "nadh dehydrogenase subnit 1", "nadh dehydrogenase ubiquinone 1 alpha",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1199 "nadh dehydrogenase subunit i");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1200 $synonym{$_} = "mtnd2" foreach ("ndh-u1", "mtnd2", "nadh dehydrogenase 2", "nadh dehydrogenase, subunit 2 (complex i)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1201 "nadh-ubiquinone oxidoreductase chain 2", "nd2", "mt-nd2", "urf2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1202 "nadh dehydrogenase subunit 2", "nd2", "nadh2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1203 "nadh-ubiquinone oxidoreductase chain 2", "nadh dehydrogenase 2", "nadh subunit 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1204 "nadh-ubiquinone oxidoreductase subunit 2", "nadh dehydrogenase subnuit 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1205 "nadh deydrogenase subunit 2", "nadh2", "nadh-ubiquinone oxidoreductase subunit ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1206 "nd2", "nadh2 protein", "nadh dehydrogenase subunit 2 (nd2)", "nadh dehydrogenase subunit ii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1207 "nadh dehydrogenase ii", "nadh dehydrogense 2", "nadh dehydrogenase subunit ii");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1208 $synonym{$_} = "mtnd3" foreach ("mtnd3", "ndh-u3", "nadh dehydrogenase 3", "nd3", "nadh3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1209 "nadh-ubiquinone oxidoreductase chain 3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1210 "nadh-ubiquinone/plastoquinone oxidoreductase chain 3", "mt-md3", "urf3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1211 "nadh dehydrogenase subunit 3", "nd3", "nadh3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1212 "nadh-ubiquinone oxidoreductase chain 3", "nadh subunit 3", "nadh3", "nadh3 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1213 "nadh-ubiquinone oxidoreductase subunit 3", "nadh dehydrogenase subnuit 3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1214 "nadh-ubiquinone oxidoreductase subunit iii", "nadh3 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1215 "nadh dehydrogenase subunit 3 (nd3)", "nadh dehydrogenase subunit iii", "nadh dehydrogenase iii",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1216 "nadh dehydrogenase subunit iii");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1217 $synonym{$_} = "mtnd4" foreach ("mtnd4", "ndh-u4", "nadh dehydrogenase 4", "nadh-ubiquinone oxidoreductase chain 4", "mt-nd4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1218 "nd4", "urf4", "nadh:ubiquinone oxidoreductase subunit 4 (chain m)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1219 "nadh dehydrogenase subunit 4", "nd4", "nadh4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1220 "nadh-ubiquinone oxidoreductase chain 4", "nadh subunit 4", "nadh4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1221 "nadh-ubiquinone oxidoreductase subunit 4", "nadh dehydrogenase subunit4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1222 "nadh-ubiquinone oxidoreductase subunit iv", "nadh4 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1223 "nadh dehydrogenase subunit 4 (nd4)", "nadh dehydrogenase subunit iv", "nadh dehydrogenase iv",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1224 "nadh dehydrogenase subunit iv");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1225 $synonym{$_} = "mtnd4l" foreach ("mtnd4l", "ndh-u4l", "nadh dehydrogenase 4l", "nadh dehydrogenase, subunit 4l (complex i)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1226 "nadh-ubiquinone oxidoreductase chain 4l", "nd4l", "mt-md4l", "urf4l",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1227 "nadh dehydrogenase subunit 4l", "nd4l", "nadh4l", "nadh-ubiquinone oxidoreductase chain 4l",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1228 "nadh subunit 4l", "nadh4l", "nadh-ubiquinone oxidoreductase subunit 4l",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1229 "nadh dehydrogenase subunit 4 l", "nadh4l protein", "nadh dehydrogenase subunit 4l (nd4l)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1230 "nadh dehydrogenase subunit ivl", "nadh dehydrogenase ivl", "nadh dehydrogenase subunit ivl");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1231 $synonym{$_} = "mtnd5" foreach ("mtnd5", "ndh-u5", "nadh dehydrogenase 5", "nd5", "nadh-ubiquinone oxidoreductase chain 5",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1232 "mt-nd5", "urf5", "nadh dehydrogenase subunit 5", "nadh5", "nd5",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1233 "nadh dehydrogenase-5", "nadh-ubiquinone oxidoreductase chain 5",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1234 "nadh dehydrogenase 5", "nadh 5", "nadh subunit 5",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1235 "nadh-ubiquinone oxidoreductase subunit 5", "nadh5", "nadh-dehydrogenase subunit 5",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1236 "nadh-dehydrogenase subunit 5, nd5", "nadh-ubiquinone oxidoreductase subunit v",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1237 "nadh5 protein", "nadh dehydrogenase subunit 5 (nd5)", "nadh dehydrogenase subunit v",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1238 "nadh dehydrogenase v", "nadh dehydrogenase subunit v");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1239 $synonym{$_} = "mtnd6" foreach ("mtnd6", "ndh-u6", "nadh dehydrogenase 6", "nd6", "nadh6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1240 "nadh-ubiquinone oxidoreductase chain 6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1241 "nadh-ubiquinone/plastoquinone oxidoreductase chain 6", "mt-md6", "urf6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1242 "nadh dehydrogenase subunit 6", "nd6", "nadh6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1243 "nadh-ubiquinone oxidoreductase chain 6", "nadh subunit 6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1244 "nadh-ubiquinone oxidoreductase subunit 6", "nadh dehydrogenase 6", "nadh6",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1245 "nadh-ubiquinone oxidoreductase subunit vi", "nadh6 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1246 "nadh dehydrogenase subunit 6 (nd6)", "nadh dehydrogenase subunit vi", "nadh dehydrogenase vi",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1247 "nadh dehydrogenase subunit vi");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1248 $synonym{$_} = "mtrnr1" foreach ("mtrnr1", "mt-rnr1", "12s rna", "12s rrna", "12s ribosomal rna", "12s rrna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1249 "small subunit ribosomal rna", "12 rrna", "12 s ribosomal rna", "12s rrna, mtssu rrna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1250
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1251 "12s rrna gene", "12s small subunit ribosomal rna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1252 "mitochondrial ssu ribosomal rna", "rrns",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1253 "ssu ribosomal rna", "12s","12s_ribosmal_rna", "s-rrna", "srrna", "ssu");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1254 $synonym{$_} = "mtrnr2" foreach ("16s", "16s_ribosomal_rna","l-rrna","lrrna","lsu", "mtrnr2", "mt-rnr2", "16s rrna", "16s rna", "16srrna", "16s ribosomal rna", "16s rrna", "rrna large subunit",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1255 "large subunit ribosomal rna", "16 s ribosomal rna", "16s mitochondrial ribosomal rna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1256
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1257 "mitochondrial 16s ribosomal rna", "16s large subunit ribosomal
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1258 rna", "16s_ribosmal_rna","rrnl");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1259
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1260 # Nuclear genes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1261 $synonym{$_} = "adora3" foreach ("adora3", "adenosine a3 receptor", "a3ar", "ara3", "gpcr2", "adora3", "adenosine a3 receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1262 "a3 adenosine receptor", "adenosine-3 receptor");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1263 $synonym{$_} = "adra2b" foreach ("adra2b", "adrenergic, alpha-2b-, receptor", "adra2l1", "adrarl1", "adra2rl1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1264 "alpha-2b adrenergic receptor", "alpha-2b adrenoceptor", "subtype c2", "[a]2b", "adra-2b",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1265 "alpha2-c2", "alpha2b", "subtype alpha2-c2", "alpha-2-adrenergic receptor-like 1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1266 "alpha adrenergic receptor 2b", "adra2b", "alpha 2b adrenergic receptor", "adra2b",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1267 "alpha adrenergic receptor subtype 2b", "alpha-2b-adrenergic receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1268 "alpha adrenergic receptor, subtype 2b", "alpha-2b adrenergic receptor", "alpha-2b adrenoceptor");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1269 $synonym{$_} = "adrb2" foreach ("adrb2", "adrenergic, beta-2-, receptor, surface", "adrb2r", "adrbr", "beta-2 adrenergic receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1270 "b2ar", "bar", "beta-2 adrenoceptor", "catecholamine receptor", "adrb-2", "badm", "beta 2-ar",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1271 "gpcr7", "adrb2", "beta-2 adrenergic receptor", "beta-2-adrenergic receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1272 "adrenergic receptor beta 2", "beta 2 adrenergic receptor", "beta2-adrenergic receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1273 "beta2 adrenergic receptor");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1274 $synonym{$_} = "apob" foreach ("apob", "apolipoprotein b", "ag(x) antigen", "apolipoprotein b-100 precursor", "apo b-100",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1275 "apolipoproteinb-48", "apo b-48", "fldb", "apob-100", "apolipoprotein b", "apolipoprotein b 100",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1276 "apob", "apolipoprotein b-100", "apob");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1277 $synonym{$_} = "app" foreach ("app", "amyloid beta (a4) precursor protein", "protease nexin-ii, alzheimer disease", "ad1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1278 "human mrna for amyloid a4 precursor of alzheimer's disease", "appi", "beta-amyloid protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1279 "beta-app", "a-beta", "a4", "cvap", "aaa", "abeta", "amyloid beta-peptide",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1280 "amyloid beta (a4) precursor protein", "adap", "appican", "betaapp", "app",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1281 "amyloid beta precursor protein", "amyloid precursor protein", "amyloid beta precursor b1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1282 "beta amyloid protein precursor", "beta-a4");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1283 $synonym{$_} = "atp7a" foreach ("atp7a", "atpase, cu++ transporting, alpha polypeptide (menkes syndrome)", "mnk",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1284 "copper-transporting atpase 1", "copper pump 1", "menkes disease-associated protein", "mc1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1285 "copper binding p-type atpase 7a", "blo", "blotchy", "br", "brindled", "menkes protein", "mo",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1286 "mottled", "mk", "ohs", "atp7a", "copper transporting atpase",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1287 "atpase, cu++ transporting, alpha polypeptide", "menkes syndrome protein");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1288 $synonym{$_} = "bdnf" foreach ("bdnf", "brain-derived neurotrophic factor", "brain-derived neurotrophic factor precursor", "bdnf",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1289 "brain-derived neurotrophic factor", "brain derived neurotrophic factor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1290 "brain-derived neurotrophic factor mature peptide", "bdnf");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1291 $synonym{$_} = "bmi1" foreach ("bmi1", "b lymphoma mo-mlv insertion region (mouse)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1292 "murine leukemia viral (bmi-1) oncogene homolog", "polycomb complex protein bmi-1", "rnf51",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1293 "mgc12685", "oncogene bmi-1", "bmi1", "bmi-1", "bmi-1 protein", "oncoprotein bmi-1");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1294 $synonym{$_} = "brca1" foreach ("brca1", "breast cancer 1, early onset", "pscp", "papillary serous carcinoma of the peritoneum",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1295 "breast cancer type 1 susceptibility protein", "breast and ovarian cancer susceptibility gene",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1296 "rnf53", "breast-ovarian cancer, included", "brca1", "brca1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1297 "breast and ovarian cancer susceptibility protein", "brca1 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1298 "breast and ovarian cancer susceptibility");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1299 $synonym{$_} = "chrna1" foreach ("chrna1", "cholinergic receptor, nicotinic, alpha polypeptide 1 (muscle)", "chrna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1300 "acetylcholine receptor protein, alpha chain precursor", "achra", "achr-1", "acra", "chrna1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1301 "nicotinic cholinergic receptor alpha polypeptide", "chrna1", "achr", "achr prepeptide", "chrna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1302 "neuronal nicotinic acetylcholine receptor alpha", "nicotinic acetylcholine recepter alpha-subunit");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1303 $synonym{$_} = "cftr" foreach ("cftr", "cystic fibrosis transmembrane conductance regulator, atp-binding cassette (sub-family c,
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1304 member 7)", "mrp7", "cf", "abcc7", "abc35", "cftr", "cystic fibrosis transmembrane conductance",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1305 "cftr chloride channel");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1306 $synonym{$_} = "cnr1" foreach ("cnr1", "cannabinoid receptor 1 (brain)", "cnr", "cb1", "cb-r", "cann6", "cb>1<", "cb1k5", "cb1a",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1307 "central cannabinoid receptor", "cnr1", "cannabinoid receptor 1", "cb1", "cb1 cannabinoid receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1308 "cb1 cannabinoid receptor", "cbr");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1309 $synonym{$_} = "crem" foreach ("crem", "camp responsive modulator", "crea", "camp-responsive element modulator, alpha isoform",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1310 "crem", "camp responsive element moderator", "camp responsive element modulator",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1311 "camp-responsive element moderator");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1312 $synonym{$_} = "edg1" foreach ("edg1", "endothelial differentiation, sphingolipid g-protein-coupled receptor, 1", "d1s3362", "edg-1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1313 "1pb1", "s1p1", "ecgf1", "chedg1", "sphingosine 1-phosphate receptor edg1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1314 "g protein-coupled sphingolipid receptor", "edg1");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1315 $synonym{$_} = "ghr" foreach ("ghr", "growth hormone receptor", "growth hormone receptor precursor", "gh receptor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1316 "serum binding protein", "growth hormone receptor", "ghr", "growth hormone receptor precursor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1317 "ghr", "bovine growth hormone receptor", "mature growth hormone receptor");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1318 $synonym{$_} = "pgk1" foreach ("pgk1", "phosphoglycerate kinase 1", "pgk-1", "primer recognition protein 2", "prp 2", "pgka",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1319 "pgk-1", "pgk-1", "phosphoglycerate kinase 1");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1320 $synonym{$_} = "plcb4" foreach ("plcb4", "phospholipase c, beta 4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1321 "1-phosphatidylinositol-4,5-bisphosphate phosphodiesterase beta 4", "plc-beta-4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1322 "phospholipase c-beta-4", "plcb4", "phospholipase c beta 4");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1323 $synonym{$_} = "pnoc" foreach ("pnoc", "prepronociceptin", "propronociceptin", "nociceptin precursor", "orphanin fq", "ppnoc", "ofq",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1324 "n/ofq", "n23k", "npnc1", "ofq/n", "proorphanin", "pnoc", "prepronociceptin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1325 "nociceptin/orphanin fq precursor");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1326 $synonym{$_} = "prkci" foreach ("prkci", "protein kinase c, iota", "pkci", "dxs1179e", "npkc-iota", "protein kinase c iota",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1327 "prkci");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1328 $synonym{$_} = "prnp" foreach ("prnp", "(prion protein (p27-30) (creutzfeld-jakob disease, gerstmann-strausler-scheinker syndrome, fatal familial insomnia)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1329 "cjd", "major prion protein precursor", "prp", "prp27-30", "prp33-35c", "ascr", "prip", "gss",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1330 "prn-i", "prn-p", "prpc", "prpsc", "sinc", "mgc26679", "cd230 antigen", "prion-related protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1331 "prion protein", "prp", "prnp", "prion protein precursor", "prion protein", "prnp", "prp",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1332 "prion protein prp", "prion protein precursor prp", "prp", "prp", "prp gene", "greater kudu prp",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1333 "major prion protein", "prion protein variant 110p", "prion protein variant 143r",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1334 "prion protein variant 240s", "prion protein variant 37v", "prion protein variant 37v/240s",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1335 "prion protein, prp", "prnp", "prmp");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1336 $synonym{$_} = "rag1" foreach ("rag1", "recombination activating gene 1", "v(d)j recombination activating protein 1", "rag-1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1337 "recombination activating protein 1", "rag1", "rag-1", "recombination activating gene-1",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1338 "recombination activating gene 1", "recombination-activating gene 1", "rag1", "rag 1", "rag1");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1339 $synonym{$_} = "rag2" foreach ("rag2", "recombination activator protein 2", "recombination activating protein 2", "rag-2", "rag 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1340 "recombination activating protein", "recombination activating gene-2", "rag2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1341 "recombination activating protein 2", "rag2", "rag2", "rag2", "rag-2", "rag-2 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1342 "recombinase activating gene 2", "recombination activating protein 2");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1343 $synonym{$_} = "rbp3" foreach ("rbp3", "retinol-binding protein 3, interstitial", "interphotoreceptor retinoid-binding protein precursor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1344 "irbp", "interstitial retinol-binding protein", "rbp-3", "interphotoreceptor retinoid binding protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1345 "irbp", "irbp", "interphotoreceptor retinoid-binding protein", "interphotorecepter retinoid binding protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1346 "interphotoreceptor retinoid binding protein (irbp)", "irbp mrna");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1347 $synonym{$_} = "tnf" foreach ("tnf", "tumor necrosis factor (tnf superfamily, member 2)", "tnfa", "dif", "tnfsf2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1348 "tumor necrosis factor (cachectin)", "tumor necrosis factor, alpha (cachectin)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1349 "tumor necrosis factor", "tnf-alpha", "cachectin", "tnfsf1a", "apc1 protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1350 "tnf, monocyte-derived", "tnf, macrophage-derived", "tnf superfamily, member 2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1351 "tumor necrosis factor-alpha", "tumor necrosis factor alpha", "tnfa", "tnf-alpha",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1352 "tumor necrosis factor alpha precursor", "tnfa", "tnfa", "tnfalpha", "tumour necrosis factor alpha",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1353 "tnf alpha", "tnf-a", "tumor necrosis factor alpha, tnf-alpha", "bovine tumor necrosis factor alpha",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1354 "tumor necrosis factor alpha (cachetin)", "tumor necrosis factor-alpha precursor");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1355 $synonym{$_} = "tp53" foreach ("tp53", "tumor protein p53 (li-fraumeni syndrome)", "p53", "cellular tumor antigen p53",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1356 "phosphoprotein p53", "trp53", "transformation related protein 53", "p53", "p53 protein", "p53",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1357 "tp53", "tumor suppressor p53", "53 kda phosphoprotein", "insulin recptor substrate p53 short form",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1358 "p53 gene product", "p53 tumor suppressor gene", "p53 tumor suppressor protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1359 "tumor suppressor p53 phosphoprotein");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1360 $synonym{$_} = "ttr" foreach ("ttr", "transthyretin (prealbumin, amyloidosis type i)", "palb", "tthy", "tbpa", "attr",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1361 "transthyretin", "transthyretin precursor", "transthyretin subunit", "ttr", "ttr");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1362 $synonym{$_} = "tyr" foreach ("tyr", "tyrosinase (oculocutaneous albinism ia)", "ocaia", "monophenol monooxygenase",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1363 "tumor rejection antigen ab", "sk29-ab", "lb24-ab", "albino", "c", "skc35", "oca1a", "tyrosinase",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1364 "tyr", "tyrosinase precursor", "truncated tyrosinase", "tyr", "tyr");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1365 $synonym{$_} = "vwf" foreach ("vwf", "von willebrand factor", "f8vwf", "coagulation factor viii", "von willebrand factor", "vwf",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1366 "von willebrand factor", "vwf", "vwf", "vwf", "von willebrand factor, vwf", "von willebrand factor precursor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1367 "von willebrand factor precursor", "wf");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1368 $synonym{$_} = "zfx" foreach ("zfx", "zinc finger protein, x-linked", "zinc finger x-chromosomal protein", "zfx", "zinc finger protein zfx",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1369 "zfx", "zinc finger protein zfx", "x-linked zinc finger protein zfx", "zfx", "x-linked zinc finger protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1370 "zinc finger x-linked protein 1", "zinc finger x-linked protein 2", "zinc finger protein x linked",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1371 "zfx1", "zfx-1", "zfx2", "zfx-2","zfx protein mrna", "zinc finger protein zfx isoform 4",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1372 "zfx product, isoform 1", "zfx product, isoform 2", "zfx product, isoform 3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1373 "zfx product, isoform 4", "zfx protein");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1374 $synonym{$_} = "zfy" foreach ("zfy", "zinc finger protein, y-linked", "zinc finger y-chromosomal protein", "zfy", "zfy",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1375 "zinc finger protein zfy", "zinc finger protein zfy", "y-linked zinc finger protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1376 "y-linked zinc finger protein zfy", "zinc finger protein y linked",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1377 "y-linked zinc finger protein 2", "y-linked zinc finger protein 1", "zfy1", "zfy-1", "zfy2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1378 "zfy-2");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1379
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1380 # Additional genes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1381 $synonym{$_} = "18s_rrna" foreach ("18s rrna", "18s_rrna", "18s ribosomal rna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1382 "18 rrna", "18 s ribosomal rna", "18 s ribosomal rna", "18s small subunit ribosomal rna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1383 "18s_ribsomal_rna", "18s_ribosomal_rna_a-type", "18s_rna_gene", "18s_rrna", "nuclear_18s_ribosomal_rna");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1384
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1385 $synonym{$_} = "28s_rrna" foreach ("28s_large_subunit_ribosomal_rna", "28s rrna", "28s_rrna", "28s ribosomal rna", "28 rrna", "put. 28S ribosomal RNA", "28 s ribosomal rna", "28 s ribosomal rna",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1386 "rrna-28s_rrna-related", "28s ribosomal rna v region");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1387 $synonym{$_} = "5.8s" foreach ("5.8s_ribosomal_rna", "5.8s_rrna", "5s_ribosomal_rna","5s_rrna","5.8s_rna_gene");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1388 $synonym{$_} = "mtcyb" foreach ("cob", "mtcyb", "cyt b", "cytb", "cytochrome b", "cytochrome b", "cytb", "cyt b", "cytb", "cytb",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1389 "'cytochrome b'", "cytochrome-b", "skeletal muscle cytochrome b", "cyt-b", "cyt.b", "cyto b",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1390 "cytob", "cytb1", "cytb2", "cytb3", "cytb4", "cytochrome b light strand", "cyt-b", "cytob",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1391 "cyt.b", "cytb", "cytochrome b", "mitochondrial cytochrome b", "cytochrome b protein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1392 "duodenal cytochrome b");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1393 $synonym{$_} = "alpha_lactalbumin" foreach ("alpha lactalbumin", "alpha_lactalbumin", "alpha-lactalbumin", "alpla lactalbumin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1394 "alpah lactalbumin", "a-lacta", "a-lacta", "lactalbumin, alpha-", "lactalbumin, alpha",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1395 "mature alpha-lactalbumin");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1396 $synonym{$_} = "alpha-s1-casein" foreach ("alpha-s1-casein", "as1-casein", "alpha s1 casein", "alpha s1-casein b",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1397 "alpha s1 casein", "alpha s1-casein", "alpha(s1) casein", "alpha-s1 casein", "alpha-s1 casein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1398 "alpha-s1-casein", "alpha-s1-casein mrna", "alphas1-casein", "as1-casein", "alfas1-casein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1399 "alpha-sl casein", "casein alpha s1");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1400 $synonym{$_} = "alpha-s2-casein" foreach ("alpha-s2-casein", "as2-casein", "alpha s2 casein", "alpha s2-casein b",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1401 "alpha s2 casein", "alpha s2-casein", "alpha(s2) casein", "alpha-s2 casein", "alpha-s2 casein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1402 "alpha-s2-casein", "alpha-s2-casein mrna", "alphas2-casein", "as2-casein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1403 "alpha s2a casein (aa 1 to 165)", "alpha s2a casein (aa 1 to 167)", "alph as2-casein",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1404 "alpha(s2)-casein");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1405 $synonym{$_} = "b-casein" foreach ("b-casein", "beta casein", "beta-casein", "beta-casein (aa 1 - 213)", "beta-casein a3",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1406 "beta-casein precursor", "beta casein precursor", "beta-casein variant a2",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1407 "beta-casein variant i", "mat. beta-casein", "casein beta", "casein b", "casein b (aa 1-183) pgpk48",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1408 "mature beta-casein (aa 1 to 226)");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1409 $synonym{$_} = "k-casein" foreach ("k-casein", "kappa casein", "kappa-casein", "kappa-casein precursor", "kappa casein precursor",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1410 "casein kappa", "kappa-cas", "kappa-casein long form", "kappa-casein mature peptide",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1411 "kappa-casein short form");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1412 $synonym{$_} = "sry" foreach ("sry", "sex determining factor sry", "sex determining region y protein", "sex-determining factor sry",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1413 "sex-determining region y", "sry gene", "sry protein", "sex region of y chromosome");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1414 $synonym{$_} = "c-mos" foreach ("c-mos", "oocyte maturation factor mos", "mos", "mos protein");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1415 $synonym{$_} = "cryaa" foreach ("cryaa", "crya1", "alpha a-crystallin","alpha a-crystallin", "alphaa-crystallin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1416 "alpha-a crystallin chain", "alpha-a-crystallin", "crystallin, alpha a", "alphaa-crystallin (crya1)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1417 "alpha a crystallin");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1418 $synonym{$_} = "cryab" foreach ("cryab", "crya2", "alpha b-crystallin","alpha b-crystallin", "alphab-crystallin",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1419 "alpha-b crystallin chain", "alpha-b-crystallin", "crystallin, alpha b", "alphab-crystallin (crya2)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1420 "alpha b crystallin");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1421 $synonym{$_} = "t-cell_receptor_beta" foreach ("t-cell receptor beta", "t cell receptor beta", "t-cell receptor beta chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1422 "t cell receptor beta chain", "t-cell receptor beta chain variable region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1423 "t cell receptor beta chain variable region", "t-cell receptor beta chain variable",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1424 "t cell receptor beta chain variable", "t-cell receptor variable beta chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1425 "t cell receptor variable beta chain", "t-cell receptor beta-chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1426 "t cell receptor beta-chain", "t-cell receptor beta chain vj region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1427 "t cell receptor beta chain vj region", "t-cell receptor beta chain v-d-j region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1428 "t cell receptor beta chain v-d-j region", "t-cell receptor beta chain variable segment",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1429 "t cell receptor beta chain variable segment", "t-cell receptor beta chain constant region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1430 "t cell receptor beta chain constant region", "t-cell receptor v beta gene",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1431 "t cell receptor v beta gene", "t-cell receptor-beta chain", "t cell receptor-beta chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1432 "t cell receptor bata chain", "t-cell receptor beta-chain", "t cell receptor beta-chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1433 "t-cell receptor beta chain (v-d-j-c)", "t cell receptor beta chain (v-d-j-c)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1434 "t-cell receptor beta-chain v region", "t cell receptor beta-chain v region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1435 "t-cell receptor v region beta-chain", "t cell receptor v region beta-chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1436 "t-cell receptor beta chain vdj region", "t cell receptor beta chain vdj region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1437 "t-cell receptor beta chain v-region", "t cell receptor beta chain v-region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1438 "t-cell receptor v-beta", "t cell receptor v-beta", "t-cell_receptor_beta");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1439 $synonym{$_} = "t-cell_receptor_alpha" foreach ("t-cell receptor alpha", "t cell receptor alpha", "t-cell receptor alpha chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1440 "t cell receptor alpha chain", "t-cell receptor alpha chain variable region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1441 "t cell receptor alpha chain variable region", "t-cell receptor alpha chain variable",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1442 "t cell receptor alpha chain variable", "t-cell receptor variable alpha chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1443 "t cell receptor variable alpha chain", "t-cell receptor alpha-chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1444 "t cell receptor alpha-chain", "t-cell receptor alpha chain vj region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1445 "t cell receptor alpha chain vj region", "t-cell receptor alpha chain v-d-j region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1446 "t cell receptor alpha chain v-d-j region", "t-cell receptor alpha chain variable segment",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1447 "t cell receptor alpha chain variable segment", "t-cell receptor alpha chain constant region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1448 "t cell receptor alpha chain constant region", "t-cell receptor v alpha gene",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1449 "t cell receptor v alpha gene", "t-cell receptor-alpha chain", "t cell receptor-alpha chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1450 "t cell receptor bata chain", "t-cell receptor alpha-chain", "t cell receptor alpha-chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1451 "t-cell receptor alpha chain (v-d-j-c)", "t cell receptor alpha chain (v-d-j-c)",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1452 "t-cell receptor alpha-chain v region", "t cell receptor alpha-chain v region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1453 "t-cell receptor v region alpha-chain", "t cell receptor v region alpha-chain",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1454 "t-cell receptor alpha chain vdj region", "t cell receptor alpha chain vdj region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1455 "t-cell receptor alpha chain v-region", "t cell receptor alpha chain v-region",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1456 "t-cell receptor v-alpha", "t cell receptor v-alpha", "t-cell_receptor_alpha");
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1457
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1458 # tRNAs
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1459 my @aminoAcids = qw(gly ala val leu ile met phe trp pro ser thr cys tyr asn gln asp glu lys arg his);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1460 my %AAcodes = ('glycine' => 'gly', 'alanine' => 'ala', 'valine' => 'val', 'leucine' => 'leu', 'isoleucine' => 'ile',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1461 'methionine' => 'met', 'phenylalanine' => 'phe', 'tryptophan' => 'trp', 'proline' => 'pro',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1462 'serine' => 'ser', 'threonine' => 'thr', 'cysteine' => 'cys', 'tyrosine' => 'tyr', 'asparagine' => 'asn',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1463 'glutamine' => 'gln', 'aspartic acid' => 'asp', 'glutamic acid' => 'glu', 'lysine' => 'lys',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1464 'arginine' => 'arg', 'histidine' => 'his');
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1465 my %fullName = ('gly' => 'glycine', 'ala' => 'alanine', 'val' => 'valine', 'leu' => 'leucine', 'ile' => 'isoleucine',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1466 'met' => 'methionine', 'phe' => 'phenylalanine', 'trp' => 'tryptophan', 'pro' => 'proline',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1467 'ser' => 'serine', 'thr' => 'threonine', 'cys' => 'cysteine', 'tyr' => 'tyrosine', 'asn' => 'asparagine',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1468 'gln' => 'glutamine', 'asp' => 'aspartic acid', 'glu' => 'glutamic acid', 'lys' => 'lysine',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1469 'arg' => 'arginine', 'his' => 'histidine');
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1470
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1471 foreach my $aa (@aminoAcids)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1472 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1473 # Abbreviations
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1474 $synonym{"trna $aa"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1475 $synonym{"transfer rna $aa"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1476 $synonym{"transfer rna-$aa"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1477 $synonym{"trna($aa)"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1478 $synonym{"trna $aa gene"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1479 $synonym{"trna-$aa gene"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1480 $synonym{"$aa trna"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1481
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1482 my $dashedName = $aa."-trna";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1483 $synonym{"$dashedName"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1484
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1485 # And again for full names
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1486 $synonym{"trna $fullName{$aa}"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1487 $synonym{"transfer rna $fullName{$aa}"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1488 $synonym{"transfer rna-$fullName{$aa}"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1489 $synonym{"trna-$fullName{$aa}"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1490 $synonym{"trna($fullName{$aa})"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1491 $synonym{"trna $fullName{$aa} gene"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1492 $synonym{"trna-$fullName{$aa} gene"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1493 $synonym{"$fullName{$aa} trna"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1494
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1495 $dashedName = $fullName{$aa}."-trna";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1496 $synonym{"$dashedName"} = "trna-$aa";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1497 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1498 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1499
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1500 sub geneClean
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1501 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1502 my $dirtyGene = shift;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1503
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1504 # Initial synonymizing
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1505 $dirtyGene = lc($dirtyGene); # Only work with lower-case for synonymizing and file names
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1506 $dirtyGene = $synonym{$dirtyGene} if (defined $synonym{$dirtyGene});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1507
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1508 # Clean end of gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1509 $dirtyGene =~ s/\"$//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1510 $dirtyGene =~ s/\s+$//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1511 $dirtyGene =~ s/^\s+//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1512
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1513 # Remove punctuation that will confound file saving
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1514 $dirtyGene =~ s/^'//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1515 $dirtyGene =~ s/'$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1516 $dirtyGene =~ s/:/ /g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1517 $dirtyGene =~ s/, / /g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1518 $dirtyGene =~ s/,//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1519 $dirtyGene =~ s/\*/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1520 $dirtyGene =~ s/\$/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1521 $dirtyGene =~ s/\#/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1522 $dirtyGene =~ s/\&/and/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1523 $dirtyGene =~ s/\//-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1524 $dirtyGene =~ s/\\//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1525 $dirtyGene =~ s/\\|/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1526 $dirtyGene =~ s/;//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1527 $dirtyGene =~ s/\<//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1528 $dirtyGene =~ s/\>//g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1529 $dirtyGene =~ s/-+/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1530 $dirtyGene =~ s/\s+-/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1531 $dirtyGene =~ s/-\s+/-/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1532 $dirtyGene =~ s/^-+//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1533 $dirtyGene =~ s/`/'/;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1534
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1535 # Collapse multiple whitespace
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1536 $dirtyGene =~ s/\s+/_/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1537
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1538 # Clean up some tRNA variants (easier than specifying explicit synonyms for each tRNA)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1539 if ($dirtyGene =~ /tRNA/i)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1540 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1541 $dirtyGene =~ s/_\d+$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1542 $dirtyGene =~ s/ \d+$//;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1543 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1544
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1545 # Final synonymizing
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1546 $dirtyGene = $synonym{$dirtyGene} if (defined $synonym{$dirtyGene}); # Recheck for synonym
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1547 $dirtyGene = lc($dirtyGene); # Ensure that gene names are in lower case for further synonymizing and file saving (final safety)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1548
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1549 return $dirtyGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1550 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1551
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1552 sub geneCount
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1553 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1554 # Set local parameters
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1555 my $gbFile = shift;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1556
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1557 my (%iGene, @geneList, $speciesNum, %quickSpeciesGenePresent, %speciesCounter);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1558 my $wordBlock = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1559 my $modelBlock = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1560 my %modelSpecies;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1561 my $stripCount = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1562
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1563 print "\nCounting gene occurrences in file $gbFile to establish genes that do not meet user-defined threshold(s)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1564
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1565 # Search input file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1566 setLineBreak($gbFile);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1567 open (GENE, "<$gbFile") or die "Cannot open GenBank output file, $gbFile.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1568 my $accCount = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1569 my $modelFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1570 my $speciesFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1571 my $nameFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1572 my $countZero = time;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1573 my ($organism, $gene);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1574 while (<GENE>)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1575 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1576 next unless ($_ =~ /LOCUS/ or $_ =~ /^\s(\/gene)\|(\/product)=\"/ or $_ =~ /^\s\/organism=\"/ or $nameFlag == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1577 chomp;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1578 my $gbLine = $_;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1579
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1580 if ($gbLine =~ /LOCUS/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1581 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1582 undef %iGene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1583 undef $organism;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1584 $modelFlag = $speciesFlag = $nameFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1585 $accCount++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1586 print "\t$accCount sequences read in\n" if ($accCount == int($accCount/10000)*10000 and $verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1587 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1588 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1589
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1590 # Get organism name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1591 if ($gbLine =~ /^\s*\/organism=\"(.+)\"/)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1592 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1593 $organism = $1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1594 $organism =~ s/\s/_/g;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1595 $modelFlag = 1 if (defined $modelSpecies{$organism});
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1596 $speciesFlag = 1 if ($organism =~ /sp\./ or $organism =~ /cf\./);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1597 unless ($speciesFlag or ($modelFlag and $modelBlock))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1598 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1599 $speciesCounter{$organism}++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1600 $speciesNum++ if ($speciesCounter{$organism} == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1601 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1602 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1603 next if ($modelFlag and $modelBlock); # Entry pertains to model organism; skip parsing rest of entry
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1604
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1605 # Get gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1606 if ($gbLine =~ /^\s*(\/gene)\|(\/product)=\"(.+)/) # Get start of gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1607 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1608 $gene = $1 if ($gbLine =~ /=\"(.+)/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1609
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1610 # Check whether name wraps onto a new line
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1611 if (substr($gbLine, -1) ne "\"")
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1612 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1613 $nameFlag = 1;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1614 next;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1615 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1616 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1617 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1618 $nameFlag = 2;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1619 $gene =~ s/\"$// if ($gene =~ /\"$/);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1620 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1621 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1622
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1623 if ($nameFlag == 1) # Get continuation / end of gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1624 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1625 $gene .= $gbLine;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1626 $nameFlag = 2 if ($gbLine =~ /\"$/) # Gene name is complete
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1627 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1628
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1629 if ($gene and $nameFlag == 2) # Process complete gene
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1630 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1631 next if (not defined $organism);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1632 unless ($wordBlock and ($gene =~ /hypothetical/i or $gene =~ /open reading frame/i or $gene =~ /similar/i or $gene =~ /homolog/i or $gene =~ /putative/i or $gene =~ /unknown/i or $gene =~ /unnamed/i or $gene =~ /\d+rik/i or $gene =~ /possible/i or $gene =~ /pseudo/i))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1633 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1634 $gene = geneClean($gene); # Clean up and synonymize gene name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1635
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1636 # Increment counters
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1637 $iGene{$gene}++; # Sequence counter
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1638 $quickSequenceCount{$gene}++ if ($iGene{$gene} == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1639
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1640 $quickSpeciesGenePresent{$gene}{$organism}++; # Species-gene counter
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1641 $quickSpeciesCount{$gene}++ if ($quickSpeciesGenePresent{$gene}{$organism} == 1);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1642 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1643 $nameFlag = 0;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1644 undef $gene;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1645 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1646 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1647 close GENE;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1648 print "\n" if ($verbose);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1649 my $countTime = time - $countZero;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1650
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1651 # Print results and block genes as appropriate
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1652 open (OUT, ">$countFile") or die "Cannot open results file for quick gene count, $countFile.\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1653 if ($speciesThreshold)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1654 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1655 @geneList = sort {$quickSpeciesCount{$b} <=> $quickSpeciesCount{$a} } keys %quickSpeciesCount;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1656 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1657 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1658 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1659 @geneList = sort {$quickSequenceCount{$b} <=> $quickSequenceCount{$a} } keys %quickSequenceCount;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1660 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1661
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1662 printf OUT "Searching $accCount sequences took $countTime seconds with %s distinct genes found\n", scalar(@geneList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1663 print OUT "\tModel organisms were excluded from search\n" if ($modelBlock);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1664 print OUT "\tCommon words were excluded from search\n" if ($wordBlock);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1665 print OUT "\nGene\tNo. of sequences\tNo. of species\n\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1666
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1667 print "\tGene counts:\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1668 foreach my $entry (@geneList)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1669 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1670 print OUT "$entry\t$quickSequenceCount{$entry}\t$quickSpeciesCount{$entry}\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1671 print "\t$entry\t$quickSequenceCount{$entry}\t$quickSpeciesCount{$entry}\n" if ($debug);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1672 if ($quickSequenceCount{$entry} < $seqThreshold or $quickSpeciesCount{$entry} < $speciesThreshold)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1673 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1674 $geneStatus{$entry} = "rejected";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1675 push @rejectList, $entry;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1676 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1677 else
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1678 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1679 $stripCount++;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1680 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1681 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1682 close OUT;
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1683
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1684 printf "\tSearching $accCount sequences took $countTime seconds with %s distinct genes found; results written to $countFile\n", scalar(@geneList);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1685 if (not $stripCount)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1686 {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1687 print "No genes were present more than the threshold value(s)\n";
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1688 exit(0);
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1689 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1690 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1691
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1692 # Version history
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1693 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1694 # v2.0 (July 25, 2005)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1695 # - improved and more complete parsing of GenBank files (esp. of gene names or CDs covering multiple lines,
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1696 # join statements and other multi-segment entries, detection of pseudogenes)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1697 # - added ability to mine input file for all gene content
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1698 # - added ability to strip only those genes present for 1) a minimum number of sequences and/or species and
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1699 # 2) species with valid names only
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1700 # - added (incomplete) gene synonymy list (initially compiled by Robin Beck)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1701 # - minor bug fixes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1702 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1703 # v1.0.1a (April 29, 2004)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1704 # - added automatic detection of line breaks
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1705 # - minor bug fixes
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1706 #
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1707 # v1.0 (April 30, 2002)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1708 # - initial release

Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics

annotate getdata/genbankstrip.pl @ 0:5b9a38ec4a39 draft default tip