#!/usr/perl/perl580/bin/perl

open (IN, "$ARGV[0]"); 
open (IN2, "$ARGV[1]"); 
open (OUT, ">$ARGV[0].ok");

##This script will generate the "in" format to my clusterization algorithm.


##This first loop will get the gene IDs and print them in the output file;
print OUT ("patient\tclass");

while (<IN2>) {
	$line = $_;
	chomp $line;
#	@fields = split (/\s:\s/, $line);
	@fields = split (/\t/, $line);
	if ($fields[0] eq '') {
		next;
	}
	print OUT ("\t$fields[0]");
#	print ("\t$fields[0]");
}
print OUT ("\n");

@fileds = ();

$i = 0;
while (<IN>) {
	$line = $_;
	chomp $line;
	$line =~ s/\r//g; ##some lina have carriage return, I *really* don't know why...
	@fields = split (/,/, $line);
	if ($fields[0] eq '') {
		next;
	}
	$class_id = pop @fields;
	print OUT ("entity_$i\t");
	print OUT ("$class_id\t");
	$line = join ("\t", @fields);
	print OUT $line;
	print OUT ("\n");
#	$a = <STDIN>;
	print ("$i\n");
	$i++;
}