Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics

annotate getdata/generate_from_phylota.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 $infile = $ARGV[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
7 my $database = $ARGV[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
8 my $outfile = $ARGV[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
9 my $treefile = $ARGV[3];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
10 my $phytabfile = $ARGV[4];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
11
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
12 open(IN, "$infile") or exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
13 open(OUT, ">$outfile") or exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
14 open(TREES, ">$treefile") or exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
15 open(DATA, "<$database") or exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
16 my @data=<DATA>;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
17 close(DATA);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
18 my @foundtrees;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
19 my @alltrees;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
20 while(<IN>){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
21 my $line = $_;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
22 chomp($line);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
23 $line =~ s/ /_/g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
24 print "Finding trees with $line ....";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
25 @foundtrees = grep /$line/, @data;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
26 my $numtrees = scalar @foundtrees;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
27 print " $numtrees Trees\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
28 if($numtrees == 0){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
29 my @genus = split(/_/,$line);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
30 @foundtrees = grep /$genus[0]/, @data;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
31 print "\tTrying genus $genus[0]";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
32 my $numtrees = scalar @foundtrees;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
33 print " $numtrees genus Trees\n";
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 push(@alltrees,@foundtrees);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
36 }
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 @alltrees = uniq(@alltrees);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
39 print TREES @alltrees;
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 #get fasta files for trees
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
42 for(my $i=0;$i < @alltrees; $i++){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
43 my @tablines = split(/\t/,$alltrees[$i]);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
44 my @tici = split(/_/, $tablines[0]);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
45 my $ti = $tici[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
46 my $ci = $tici[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
47 my $addstring = $ti.$ci."_";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
48 $ti =~ s/ti//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
49 $ci =~ s/ci//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
50 my $fastafile = getfastafromphylota($ci,$ti);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
51 #Add TI_CI_ to each fastaheader
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
52 $fastafile =~ s/\>/\>$addstring/g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
53 print OUT $fastafile;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
54 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
55 close(IN);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
56 close(OUT);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
57 close(TREES);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
58
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
59 #Now convert fasta file to phytab file and write
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
60 open(PHYTAB, ">$phytabfile") or exit;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
61 # open infile fasta file
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
62 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
63 my $total=0;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
64 # grab sequence object
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
65 while (my $seq = $in_obj->next_seq() ) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
66 my $seq_obj = $in_obj;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
67 my $sequenceid = $seq->id;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
68 my $species_name = $seq->desc;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
69 my $fullheader = $sequenceid." ".$species_name;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
70 my $sequence = $seq->seq;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
71 my @header = split(/_/, $fullheader);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
72 my $cluster = $header[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
73 my $seqgi = $header[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
74 $seqgi =~ s/gi//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
75 my $seqti = $header[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
76 $seqti =~ s/ti//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
77 my $seqsp = $header[3];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
78 $seqsp = cleansp($seqsp);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
79 print "Writing phytab for $seqsp\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
80 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
81 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
82 close(PHYTAB);
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 #remove duplicate lines (trees)
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
85 sub uniq {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
86 my %seen = ();
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
87 my @r = ();
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
88 foreach my $a (@_) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
89 unless ($seen{$a}) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
90 push @r, $a;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
91 $seen{$a} = 1;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
92 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
93 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
94 return @r;
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 cleansp
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 $seqsp = shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
99 $seqsp =~ s/ /_/g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
100 $seqsp =~ s/\.//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
101 $seqsp =~ s/\'//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
102 $seqsp =~ s/\-//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
103 return($seqsp);
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 sub getfastafromphylota
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
106 {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
107 my $ci=shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
108 my $ti=shift;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
109
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
110 print "Writing: CI:$ci TI:$ti\n";
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 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
113 my $content = get $url;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
114 die "Couldn't get $url" unless defined $content;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
115 $content =~ s/\<html\>\<pre\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
116 $content =~ s/\<\/html\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
117 $content =~ s/\<\/pre\>//;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
118 return($content);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
119 }