#! usr/bin/perl -w

use warnings;
use strict;

use Bio::Seq;
use Bio::SeqIO;

#print "What is the filepath of the input file?\n";
my $file = $ARGV[0];
#print "What is the name of the output file?\n";
my $output = $ARGV[1];
my $trash = $ARGV[2];

my $seq_in = Bio::SeqIO->new(-file => $file, -format => "fasta");


open OUTPUT, ">$output";
open TRASH, ">$trash";

my $temp_seq_counter = 0;
my $match_counter = 0;
my $seq;
my $tag_id;
my $position;
my $tag_name;

while(my $seq_obj = $seq_in->next_seq){

	$temp_seq_counter += 1;
	my $temp_seq = $seq_obj->seq;
	my $temp_seq_name = $seq_obj->id;

	my $tag_file = Bio::SeqIO->new(-file => "</home/Alan/Desktop/sequence_data/INFLUENZA_01_07_2010/MIDS.fasta", -format => "fasta");



	while(my $tag_obj = $tag_file->next_seq){

		my $tag = $tag_obj->seq;
		print "tag = ",$tag,"\n";
		my $RC_tag = reverse $tag =~ tr/ACTGactg/TGACtgac/;

		if($temp_seq =~ m/$tag/g || $temp_seq =~ m/$RC_tag/g){

			$position = pos($temp_seq);
			my $length = length($temp_seq);
			
			

				$seq = $seq_obj->subseq($position,$length);
				
				$tag_name = $tag_obj->id;

				$match_counter = 1;
		}

	}

	if($match_counter == 1){
		print OUTPUT ">",$temp_seq_name," Tag: ",$tag_name," ending at ",$position,"\n",$seq,"\n";
	}
	else{

		print TRASH ">",$temp_seq_name," ",$seq_obj->desc,"\n",$temp_seq,"\n";

	}

	$match_counter = 0;

}

print $temp_seq_counter," seqeuences were tested and ",$match_counter," seqeuences have a tag.\n";

close OUTPUT;
close TRASH;