perl/etc/crep/merge_crep_anno.pl

   1 #!/bin/env perl
   2 use strict;
   3 use warnings;
   4 #use IO::Unread qw(unread);
   5 use Term::ANSIColor qw(:constants);
   6 use Data::Dump qw(ddx);
   7
   8 die "Usage: $0 <out> <input1> [input2 ...]\n" if @ARGV < 2;
   9 my $out = shift @ARGV;
  10
  11 my $psif='./Rice.psi';
  12
  13 my (@psid,$tmp,%ANNO,%INPUTIDS);
  14 open PSI,'<',$psif or die "Error: $!\n";
  15 while (<PSI>) {
  16         next if /^#/;
  17         $tmp = (split /\t/,$_)[1];
  18         push @psid,$tmp;
  19         ++$INPUTIDS{$tmp};
  20         #print "[$tmp_line]\n";
  21 }
  22 close PSI;
  23 $tmp = $#psid + 1;
  24 warn GREEN,BOLD,$tmp,RESET," ProbeSets loaded.\n";
  25
  26 while (<>) {
  27         chomp;
  28         my @dat = split /\t/;
  29         next unless exists $dat[1];
  30         $ANNO{shift @dat} = \@dat;
  31 }
  32 $tmp = (keys %ANNO);
  33 warn GREEN,BOLD,$tmp,RESET," ProbeSets Found.\n";
  34
  35 if ($tmp < $#psid + 1) {
  36         print "Lacking:\n";
  37         for (@psid) {
  38                 unless (exists $ANNO{$_}) {
  39                         print "[$_]\n";
  40                 }
  41         }
  42         die;
  43 }
  44 if ($tmp > $#psid + 1) {
  45         print "More:\n";
  46         for (keys %ANNO) {
  47                 unless (exists $INPUTIDS{$_}) {
  48                         print "[$_]\n";
  49                 }
  50         }
  51 }
  52
  53 open O,'>',$out or die "Error: $!\n";
  54 for (@psid) {
  55         print O join("\t",$_,@{$ANNO{$_}}),"\n";
  56 }
  57 close O;
  58
  59 __END__
  60 perl merge_crep_anno.pl crep_anno_merged.tsv crep_anno_tsv.all.txt crep_anno_tsv.txt*
  61 wc -l crep_anno_merged.tsv