Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics

annotate orthologs/ucsb_hamster/translate.pl @ 0:5b9a38ec4a39 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
First commit of old repositories
author osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
date Tue, 11 Mar 2014 12:19:13 -0700
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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 use strict;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
3 use FindBin;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
4 use lib "$FindBin::Bin/lib";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
5 use Getopt::Long;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
6 use Bio::Perl;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
7 use Bio::SeqIO;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
8 use File::Copy;
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 # PROGRAMNAME: translate.pl
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
11
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
12 # AUTHOR: INGO EBERSBERGER, ingo.ebersberger@univie.ac.at
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 # PROGRAM DESCRIPTION:
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
15
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
16 # DATE: Tue May 12 14:03:34 CEST 2009
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
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
19 # DATE LAST MODIFIED:
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
20 ######################## start main #############################
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
21 my $help;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
22 my @out;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
23 my @estout;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
24 my $infile;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
25 my $trunc = 1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
26 my $outfile = "translate_tc.out";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
27 my $limit = 20; ## this sets the maximum length for the sequence identifier. If sequence identifier are
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
28 ## too long, then one can run into troubles with the parsing of the hmmsearch results.
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
29 open (LOG, ">>hamstrsearch.log") or warn "could not open logfile for writing\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
30 print LOG "##### Translation of the ESTs########";
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 my $usage = "Name:\n\ttranslate.pl\n
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
33 Synopsis:\n\ttranslate_tc5.pl [-infile=FILE] [options] [-outfile=FILE]\n
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
34 Description:\n\tThis program takes a batch fasta-file with DNA
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
35 \tsequences as an input and translates the individual DNA sequences in
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
36 \tall six reading frames.
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
37 \t-infile: provide the relative or absolute path of the infile\n
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
38 \t-outfile: provide the realtive or absolute path of the
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
39 \toutfile. Default is: translate_tc.out\n
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
40 \ttrunc: set -trunc=0 to prevent truncation of the sequence header (see below).
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
41 \t-h: prints this help-message\n
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
42 NOTE: if the seq-id (everything up to the first [[:space:]]) contains a '|' everything between the '>' and the '|' will be taken as seq-id. Otherwise, the entire seq-id will be used. You can change this behavior by setting -trunc=0\n
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
43 NOTE: the script as an automated routine to check for unique sequence names in the input file. This may lead to cases where the $trunc value is overruled and additionally part of the sequence description may be included.";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
44 ##########
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
45
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
46 GetOptions (
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
47 "h" => \$help,
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
48 "infile=s" => \$infile,
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
49 "outfile=s" => \$outfile,
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
50 "trunc=s" => \$trunc);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
51 if ($help) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
52 print "$usage";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
53 exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
54 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
55 if (-e "$outfile") {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
56 print LOG "an outfile $outfile already exists. Renaming to $outfile.old\n\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
57 my $newname = "$outfile.old";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
58 rename($outfile, $newname);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
59 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
60 ##BUG -- BIOPERL GUESSES PROTEIN FILE FORMAT WHEN AMBIGUITY CODES ARE PRESENT
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
61 ##CAUSING AN ERROR IN THE TRANLATE_6 FRAMES, WHICH INTERRUPTS ALL TRANSLATION -- THO
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
62 #below replaces read_all_sequences function to declare that sequence is DNA
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
63 #original line next
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
64 #my @seq_object = read_all_sequences($infile, 'fasta');
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 my $tempseqio;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
67 $tempseqio = Bio::SeqIO->new( '-file' => $infile, '-format' => 'fasta');
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
68 my @seq_object;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
69
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
70 while( my $seq = $tempseqio->next_seq() ) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
71 $seq->alphabet('dna');
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
72 push(@seq_object,$seq);
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 #End THO changes
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
75
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
76
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
77 ## determine whether the seq-ids are unique given the chosen value for $trunc
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
78 my ($message, $cont, $check) = &checkIds();
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
79 if ($cont == 1) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
80 ## the check for unique identifiers has failed and the programm is exiting
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
81 print LOG "$message\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
82 exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
83 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
84 else {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
85 print LOG "All sequence identifier are unique!\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
86 if ($check == 2) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
87 my $newname = "$infile.original";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
88 rename($infile, $newname);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
89 print LOG "Sequence description was needed to make seq-id unique. The original version of the infile was stored in $infile.original\n";
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 for (my $j = 0; $j < @seq_object; $j++) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
92 my $finalid = $seq_object[$j]->{finalid};
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
93 my $estseq = $seq_object[$j]->seq;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
94 my $inid = $seq_object[$j]->display_id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
95 my @all_trans = Bio::SeqUtils->translate_6frames($seq_object[$j]);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
96 for (my $i = 0; $i < @all_trans; $i++) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
97 my $count = $i+1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
98 my $id = $all_trans[$i]->display_id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
99 my $seq = $all_trans[$i]->seq;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
100 $id =~ s/$inid/$finalid/;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
101 $id =~ s/-[0-9][RF]/_RF$count.0/;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
102 push @out, ">$id\n$seq";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
103 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
104 push @estout, ">$finalid\n$estseq";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
105 if ($j%100 == 0) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
106 print "$j Sequences processed\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
107 open (OUT, ">>$outfile") or die "failed to open outfile\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
108 print OUT join "\n", @out;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
109 print OUT "\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
110 @out = qw();
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
111 close OUT;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
112 if ($check == 2) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
113 ## part of the description was added to the seq-id
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
114 open (OUT, ">>$infile");
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
115 print OUT join "\n", @estout;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
116 print OUT "\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
117 @estout = qw();
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
118 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
119 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
120 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
121 open (OUT, ">>$outfile") or die "failed to open outfile\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
122 print OUT join "\n", @out;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
123 print OUT "\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
124 @out = qw();
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
125 close OUT;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
126 if ($check == 2) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
127 ## part of the description was added to the seq-id
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
128 open (OUT, ">>$infile");
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
129 print OUT join "\n", @estout;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
130 print OUT "\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
131 close OUT;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
132 @estout = qw();
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 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
135 close LOG;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
136 exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
137 ########################## start sub ################
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
138 sub checkIds {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
139 my $message;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
140 my $check = 1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
141 my $cont = 1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
142 my $counter;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
143 ## Everything up to the first whitespace
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
144 ## in the fasta header will be taken as sequence id by bioperl. If this
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
145 ## id contains a '|' and $trunc is set to 1 (default), the ids may no longer
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
146 ## be unique. This will be checked and if necessary the id will not be truncated
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
147 ## for $check == 0, the truncated version of the id will be checked (only if $trunc == 1)
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
148 ## for $check == 1, the complete id will be checked
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
149 ## for $check == 2, the first 20 characters of the concatenated id and description
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
150 ## will be checked
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
151 if ($trunc == 1) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
152 $check = 0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
153 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
154
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
155 while ($check < 3 and $cont == 1) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
156 $cont = 0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
157 for (my $i=0; $i < @seq_object; $i++) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
158 my $id = $seq_object[$i]->display_id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
159 $id =~ s/(.{0,$limit}).*/$1/;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
160 if ($check == 0) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
161 $id =~ s/|.*//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
162 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
163 elsif ($check == 2) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
164 print "Check 2: ".$seq_object[$i]->display_id." \n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
165 $id = $id . '_' . $seq_object[$i]->desc;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
166 $id =~ s/(.{0,$limit}).*/$1/;
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 if (defined $counter->{$id}) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
169 if ($check == 0) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
170 $message = "trying next without truncating the id";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
171 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
172 elsif ($check == 1) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
173 $message = 'trying next to include sequence description';
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
174 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
175 else {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
176 $message = "Sequence identifier are not unique, using the first 20 characters. Aborting...";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
177 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
178 print LOG "sequence ids are not unique in the file $infile, $message. The offending identfier is $id\n\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
179 print "sequence ids are not unique in the file $infile, $message. The offending identfier is $id\n\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
180 $check ++;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
181 $cont = 1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
182 $counter = undef;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
183 last;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
184 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
185 else {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
186 $counter->{$id} = 1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
187 $seq_object[$i]->{finalid} = $id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
188 }
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 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
191 ## return the value of $cont. If this is 1, then the sequence id check has failed.
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
192 return($message, $cont, $check);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
193 }