lite2tg

#!/usr/bin/perl

use warnings;
use strict;
use fralib;
use Getopt::Long;
use File::Basename;
use Pod::Usage;

=head1 NAME

lite2tg

=head1 SYNOPSIS

 affymetrix2tg [options] <affymetrix-call-file>	

  -h                     help
  -o                     output file name (optional)
                         default: replaces extension of 
                         <affymetrix-call-file>	with tg
  affymetrix-call-file   Affymetrix calls file
       
 example: affymetrix2tg -o pscalare.tg brlmm.calls.txt
       
 Converts the affymetrix calls output file to a tg-file.
       
=head1 DESCRIPTION

=cut

my $help;
my $affymetrixFile;
my $tgFile;
my $headerProcessed = 0;
my $colNo;

#initialize options
Getopt::Long::Configure ('bundling');

if(!GetOptions ('h'=>\$help, 'o=s'=>\$tgFile) || scalar(@ARGV)!=1)
{
    if ($help)
    {
        pod2usage(-verbose => 2);
    }
    else
    {
        pod2usage(1);
    }
}

$affymetrixFile = $ARGV[0];

if (!defined($tgFile))
{
    my ($name, $path, $ext) = fileparse($affymetrixFile, '\..*');
	$tgFile = "$name.tg";
}

open(IN, $affymetrixFile) || die "Cannot open $affymetrixFile\n";
open(OUT, ">$tgFile") || die "Cannot open $tgFile\n";

while (<IN>)
{
	s/\r?\n?$//;
	
	if(!$headerProcessed) 
	{
		if(/^probeset_id/)  
		{
			$colNo = s/\t/\t/g + 1;
			my @fields = split('\t',$_,$colNo);

			print OUT "snp-id";

			#reads in sample ids
			for my $col (1 .. $#fields)  
			{			
				print OUT "\t$fields[$col]";			
			}
			print OUT "\n";
			$headerProcessed = 1;
		}
	}
	else
	{
		my @fields = split(/\t/);
							
		print OUT "$fields[0]";
			
		for my $col (1 .. $#fields)  
		{
			if($fields[$col]=~/^(0|1|2|-1)$/)
			{
				print OUT "\t$fields[$col]";
			}
			else
			{
				die "Unrecognized encoding in $affymetrixFile: $fields[$col]";
			}
		}
			
		print OUT "\n";
	}
}

close(OUT);
close(IN);