my $slash = "/";
if("$^O" eq "MSWin32"){
	$slash = "\\";
}
# restrict the search to a specific taxonomy
my $taxon = $ARGV[3];
$annotation .= "\t$taxon";
open(tax_file, "..".$slash."dataset".$slash."taxonomy.tab") or die "couldn't open taxonomy.tab";
my @taxR = <tax_file>;
close tax_file;
if($taxon){
	@taxR = grep { /$taxon/i } @taxR;
	for(my $e = 0; $e < scalar(@taxR); $e++){
		my @taxRR = split(/\t/, $taxR[$e]);
		$taxR[$e] = $taxRR[1];
	}
}
my %taxR = map { $_ => 1 } @taxR;
print "cyaa5 = ".$taxR{"CYAA5"};#prints cyaa5 = 1
#skipping a bunch of unrelated stuff
	open(ps_file, "..".$slash."dataset".$slash.$tempFile) or die "couldn't open $tempFile";
	while(<ps_file>){
		chomp;
		
		
		my @curLine = split(/\t/, $_);
		my $filter = 1;
		if($taxon){
			print "$curLine[2]\t$taxR{$curLine[2]}\n";#produced weird output with when run with the huge protein file will post below
                        #these commented out lines are previous attempts that work on windows but not Mac
			#$filter = $curLine[2] ~~ @taxR;
			#$filter = scalar(grep( /^$curLine[2]$/, @taxR ));
			#$filter = ( first { $_ eq $curLine[2] } @taxR );
			#print $taxR{curLine[2]}."\n";
			$filter = $taxR{$curLine[2]};
		}
		if($filter){
			checkSeq(@curLine);
		}
	}
	close ps_file;