Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics

annotate getdata/phylota_with_taxid.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 LWP::Simple;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
4 use Bio::SeqIO;
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 my $ti = $ARGV[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
7 my $outfile = $ARGV[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
8 my $phytabfile = $ARGV[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
9
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
10 open(OUT, ">$outfile") or exit;
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 my $content = getclustersfromphylota($ti);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
13 my @weblines = split(/\<\/tr\>/, $content);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
14 my @ci;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
15
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
16 #Parse html from phylota browser to retain just each ci
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
17 foreach(@weblines){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
18 if($_ =~ m/getcluster\.cgi/){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
19 chomp;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
20 $_ =~ s/\&ntype\=1\&db\=184\".+// ;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
21 $_ =~ s/(.*?)getcluster\.cgi.+cl\=// ;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
22 $_ =~ s/\<\/font\>\<\/td\>// ;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
23 chomp;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
24 $_ =~ s/^\n// ;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
25 push(@ci, $_);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
26 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
27 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
28
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
29 #get fasta files for trees
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
30 for(my $i=0;$i < @ci; $i++){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
31 my $ci = $ci[$i];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
32 my $addstring = 'ti'.$ti.'ci'.$ci.'_';
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
33 my $fastafile = getfastafromphylota($ci,$ti);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
34 #Add TI_CI_ to each fastaheader
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
35 $fastafile =~ s/\>/\>$addstring/g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
36 print OUT $fastafile;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
37 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
38 close(OUT);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
39
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
40 #Now convert fasta file to phytab file and write
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
41 open(PHYTAB, ">$phytabfile") or exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
42 # open infile fasta file
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
43 my $in_obj = Bio::SeqIO->new(-file => $outfile, '-format' =>'fasta');
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
44 my $total=0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
45 # grab sequence object
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
46 while (my $seq = $in_obj->next_seq() ) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
47 my $seq_obj = $in_obj;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
48 my $sequenceid = $seq->id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
49 my $species_name = $seq->desc;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
50 my $fullheader = $sequenceid." ".$species_name;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
51 my $sequence = $seq->seq;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
52 my @header = split(/_/, $fullheader);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
53 my $cluster = $header[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
54 my $seqgi = $header[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
55 $seqgi =~ s/gi//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
56 my $seqti = $header[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
57 $seqti =~ s/ti//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
58 my $seqsp = $header[3];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
59 $seqsp = cleansp($seqsp);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
60 print PHYTAB $seqsp."\t".$cluster."\t".$seqgi."\t".$sequence."\n";
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 close(PHYTAB);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
63
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
64
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
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
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 #sub routines
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 sub cleansp
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 $seqsp = shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
75 $seqsp =~ s/ /_/g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
76 $seqsp =~ s/\.//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
77 $seqsp =~ s/\'//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
78 $seqsp =~ s/\-//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
79 return($seqsp);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
80 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
81 sub getfastafromphylota
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
82 {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
83 my $ci=shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
84 my $ti=shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
85
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
86 #print "Writing: CI:$ci TI:$ti\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
87
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
88 my $url = 'http://phylota.net/cgi-bin/sql_getcluster_fasta.cgi?format=all&db=184&ti='.$ti.'&cl='.$ci.'&ntype=1';
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
89 my $content = get $url;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
90 die "Couldn't get $url" unless defined $content;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
91 $content =~ s/\<html\>\<pre\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
92 $content =~ s/\<\/html\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
93 $content =~ s/\<\/pre\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
94 return($content);
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 sub getclustersfromphylota
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
97 {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
98 my $ti=shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
99
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
100 #print "Writing: CI:$ci TI:$ti\n";
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 my $url = 'http://phylota.net/cgi-bin/sql_getclusterset.cgi?ti='.$ti.'&ntype=1&piflag=1&dflag=0&db=184';
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
103 my $content = get $url;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
104 die "Couldn't get $url" unless defined $content;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
105 $content =~ s/\<html\>\<pre\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
106 $content =~ s/\<\/html\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
107 $content =~ s/\<\/pre\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
108 return($content);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
109 }