#!/usr/bin/perl -w
#1_copy_both_spp.pl
use strict; use warnings;

##NB: To run this script on the file my_orthomcl_dir/groups.txt, specify it when runniong the script to add it to the @AGRV array

die "usage: 1_copy_both_spp.pl <file>\n" if (@ARGV!=1);
print "Processing file $ARGV[0]\n";

##counting groups that have a single copy in both spp
my $bothspp=0;



my $outfile ="1_copy_both_spp_output.txt";

open (my $out, ">$outfile") or die ("Problem with opening $outfile: $!");



while (my $line=<>){
	next unless($line=~m/group\d{1,4}\:\s+cct\|\w{6}\sppb\|\w{6}/);
	$bothspp++;
	my $count+= ()=($line =~ m/\|/g);
	print $out "$line \($count proteins\)\n";
}
print "There are $bothspp groups of protein families that contain a single copy in both spp\n";
close($out);

##output "There are 1468 groups of protein families that contain a single copy in both spp.