#!/usr/bin/perl -w
#unique_cct.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: unique_cct.pl <file>\n" if (@ARGV!=1);
print "Processing file $ARGV[0]\n";

##counting lines/protein groups unique to ppb
my $uniqppb=0;

my $outfile ="unique_ppb_output.txt";

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

while (my $line = <>){
	next unless ($line=~/ppb/);
	next unless ($line!~/cct/);
	$uniqppb++;
	my $count+= ()=($line =~ m/\|/g);
	print $out "$line \($count proteins\)\n";
}
print "There are $uniqppb groups of protein families unique to ppb\n";
close($out)

##output "There are 113 groups of protein families uniqe to ppb"