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

##counting groups that have a single copy in ppb and multiple copies in cct
my $bothspp=0;


my $outfile ="1_cp_ppb_x_cp_cct_output.txt";

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


while (my $line=<>){
	next unless($line=~m/cct\|\w{6}\scct\|\w{6}/);
	next unless ($line=~m/ppb/);
	next if ($line=~m/ppb\|\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 of ppb, but multiple copies of cct\n";
## "There are 7 groups of protein families that contain a single copy of ppb, but multiple copies of cct.

my %count;
my $file = shift or die "Usage: $0 FILE\n";