find_omark_contamination.pl

gistfile1.pl

#!/bin/env perl
open IN,shift;
while(<IN>){
	push @h,"$1" if />(\S+)/
}
close IN;
open IN,shift;
while(<IN>){
	@F=split;
	next unless $F[2] eq 'CDS';
	$F[9]=~s/[;"]//g;
	$c{$F[9]}=$F[0];
}

foreach $g (@h){
	# print "$g => $c{$g}\n";
	$counts{$c{$g}}++
}

while(($k,$v)=each %counts){print "$k => $v\n"}