Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics

annotate getdata/get_gb.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
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
4 #use FindBin;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
5 #use lib "$FindBin::Bin/lib";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
6 use Bio::DB::GenBank;
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
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 my $datafile = $ARGV[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
11 my $datatype = $ARGV[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
12 my $outtype = $ARGV[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
13 my $outfile = $ARGV[3];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
14 my $manual = $ARGV[4];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
15 my $mannames = $ARGV[5];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
16 my $genenames = $ARGV[6];
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 my $accessions;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
20 my @accnums;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
21 my @newnames;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
22 my $manbin=0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
23 my @genenames;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
24 my $genebin=0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
25
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
26 unless($mannames eq ''){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
27 @newnames = split(/ /,$mannames);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
28 $manbin=1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
29 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
30
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
31 unless($genenames eq ''){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
32 @genenames = split(/ /,$genenames);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
33 $genebin=1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
34 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
35
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
36 if($datafile eq 'None'){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
37 @accnums = split(/ /,$manual);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
38 # if(@accnums != @newnames && $manbin ==1 ){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
39 # die "Must have the same number of Custom Names as Accession Numbers\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
40 # }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
41 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
42 open (FILE,"<$datafile") or die "Cannot open file containing accession numbers\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
43
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
44 while (<FILE>)
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 chomp;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
47 next unless ($_);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
48 push(@accnums, $_);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
49 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
50 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
51 my $countnames = 0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
52 foreach (@accnums){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
53 #Should check input for one word per line and throw error if not, which is not done
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 $accessions = $_;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
56 chomp;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
57 if($accessions eq ""){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
58 die "Put spaces between accession numbers\n";
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 my $qry_string .= $accessions."[accession]"." ";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
61
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
62 my $GBseq;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
63 my $gb = new Bio::DB::GenBank;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
64 my $query = Bio::DB::Query::GenBank->new
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
65 (-query =>$qry_string,
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
66 -db =>$datatype);
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 my $count;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
69 my $species;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
70 my $seqio;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
71 if($outtype eq "phytab"){ #print phytab format, do not use bioperl as below.
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
72 open(OUTFILE, ">>$outfile");
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
73 if( defined ($seqio = $gb->get_Stream_by_query($query)) ){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
74 # my $seqio = $gb->get_Stream_by_query($query);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
75 while( defined ($GBseq = $seqio->next_seq )) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
76 my $sequence = $GBseq; # read a sequence object
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
77 if($manbin ==1){ #replace GenBank Names with Custom Names
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
78 $sequence->id($newnames[$countnames]);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
79 $sequence->desc('');
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
80 $species = $sequence->id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
81 $countnames++;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
82 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
83 $species = $sequence->species->binomial;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
84 $species =~ s/ /_/g ;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
85 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
86 if(@genenames > 0){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
87 if(@genenames == 1){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
88 print OUTFILE $species."\t".$genenames[0]."\t".$sequence->accession."\t".$sequence->seq."\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
89 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
90 print OUTFILE $species."\t".$genenames[$countnames-1]."\t".$sequence->accession."\t".$sequence->seq."\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
91 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
92 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
93 print OUTFILE $species."\tNone\t".$sequence->accession."\t".$sequence->seq."\n";
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 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
96 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
97 print "Did not find $accessions\n";
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 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
100 my $fh = Bio::SeqIO->newFh(-format=>$outtype, -file=>">>$outfile");
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
101
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
102 if( defined ($seqio = $gb->get_Stream_by_query($query)) ){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
103 # my $seqio = $gb->get_Stream_by_query($query);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
104 while( defined ($GBseq = $seqio->next_seq )) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
105 my $sequence = $GBseq; # read a sequence object
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
106 if($manbin ==1){ #replace GenBank Names with Custom Names
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
107 $sequence->id($newnames[$countnames]);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
108 $sequence->desc('');
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
109 $countnames++;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
110 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
111 print $fh $sequence; # write a sequence object
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
112 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
113 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
114 print "Did not find $accessions\n";
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 }
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 exit;