Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics

annotate phyloconversion/remove_phytab_dupes.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
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
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 $keeplongest=$ARGV[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
8 my $ignoregaps=$ARGV[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
9 my $uniout=$ARGV[3];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
10 my $dupout=$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 die "Cannot open $infile\n";
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 my %UniquesHash;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
15 my @DupeArray;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
16
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
17 while(<IN>){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
18 my $row = $_;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
19 chomp($row);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
20 my @column = split(/\t/, $row);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
21 my $species = $column[0];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
22 my $partition = $column[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
23 my $id = $column[2];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
24 my $sequence = $column[3];
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 if(exists $UniquesHash{$species}{$partition}){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
27 my @dupeseq = split(/\t/, $UniquesHash{$species}{$partition});
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
28 my ($savlen,$curlen);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
29 if($ignoregaps==1){
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
30 my $nogapsav = $dupeseq[1];
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
31 my $nogapcur = $sequence;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
32 $nogapsav =~ s/\-//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
33 $nogapcur =~ s/\-//g;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
34 $savlen = length($nogapsav);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
35 $curlen = length($nogapcur);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
36 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
37 $savlen = length($dupeseq[1]);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
38 $curlen = length($sequence);
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 if($curlen > $savlen && $keeplongest==1) { #current is longer so keep that one
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
41 my $oldline = $species."\t".$partition."\t".$UniquesHash{$species}{$partition}."\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
42 $UniquesHash{$species}{$partition} = "$id\t$sequence";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
43 push(@DupeArray, $oldline);
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
44 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
45 push(@DupeArray, "$species\t$partition\t$id\t$sequence\n");
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
46 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
47 }else{
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
48 $UniquesHash{$species}{$partition} = "$id\t$sequence";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
49 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
50 }
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
51
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
52 open OUT, ">".$uniout or die "Cannot open $uniout\n";
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
53 open DUPES, ">".$dupout or die "Cannot open $dupout\n";
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 print DUPES @DupeArray;
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
56 for my $spname ( keys %UniquesHash ) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
57 for my $partname ( keys %{ $UniquesHash{$spname} } ) {
5b9a38ec4a39 First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff changeset
58 print OUT "$spname\t$partname\t$UniquesHash{$spname}{$partname}\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 }