#!/usr/bin/perl -w
# A program to parse tfscan output and print CisML formatted data

# Read in header lines and get sequence data
$line = <>;
chomp $line;
($seq_name, $start, $stop) = $line =~ /of\s+(.*)\s+from\s+(\d+)\s+to\s+(\d+)/;
$sequence_length = ($stop - $start) + 1;
$line = <>;

# Read in hits and hash by pattern accession
while ($line = <>) {
    chomp $line;
    ($pattern_name, $pattern_acc, $hit_start, $hit_stop, $hit_seq) = split(/\s+/,$line);
    unless (defined($patterns{$pattern_acc})) { $patterns{$pattern_acc} = $pattern_name; }
    push(@{$hits{$pattern_acc}}, [$hit_start, $hit_stop, $hit_seq]);
}

#Print CisML header
print "<?xml version=\"1.0\"?>\n";
print "<cis-element-search\n";
print "xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n";
print "xsi:schemaLocation=\"http://zlab.bu.edu/schema/cisml cisml.xsd\"\n";
print "xmlns=\"http://zlab.bu.edu/schema/cisml\"\n";
print ">\n";

print "\t<program-name>tfscan</program-name>\n";
  
#Print CisML hits
foreach $pattern (keys %hits) {
    print "\t<pattern accession=\"$pattern\" name=\"$patterns{$pattern}\">\n";
    print "\t\t<scanned-sequence accession=\"$seq_name\" name=\"$seq_name\" length=\"$sequence_length\">\n";
    foreach $match (@{$hits{$pattern}}) {
        print "\t\t\t<matched-element start=\"$$match[0]\" stop=\"$$match[1]\"><sequence>$$match[2]</sequence></matched-element>\n";
    }
    print "\t\t</scanned-sequence>\n";
    print "\t</pattern>\n";
}

#Print end of CisML
print "</cis-element-search>\n";




=head1 NAME

tfscan2cisml.pl

=head1 SYNOPSIS

tfscan2cisml.pl

=head1 DESCRIPTION

Converts output of tfscan (http://ocgc.ca/programs/emboss/tfscan.html) to CisML XML format.
tfscan output does not provide all of the information generally included in CisML.

tfscan is part of the EMBOSS package and can be obtained from http://ocgc.ca/programs/emboss/tfscan.html

tfscan output looks like this:

TFSCAN of HSFOS from 1 to 6210

HS$CFOS_20           R08485   384   396   agttcccgtcaat
DOG$ATP1A_01         R08484   3057  3063  gacatgg
HS$CEBPA_01          R08471   4535  4540  cacgtg
HS$GPB_05            R08210   3716  3721  gtatct
...


=head1 OPTIONS



=head1 EXAMPLES

tfscan2cisml.pl tfscan_file

=head1 COPYRIGHT

# Copyright (C) 2004 by Peter M. Haverty, Trustees of Boston University

=head1 AUTHOR

Peter M. Haverty phaverty@bu.edu

=head1 SEE ALSO



=head1 REVISION HISTORY

	$Id: tfscan2cisml.pl,v 1.3 2004/02/22 21:41:37 phaverty Exp $

	$Log: tfscan2cisml.pl,v $
	Revision 1.3  2004/02/22 21:41:37  phaverty
	dunno
	
	Revision 1.2  2004/02/18 15:26:56  phaverty
	Changed sequence in matched-element to a sequence element.
	
	Revision 1.1  2004/02/18 02:41:28  phaverty
	Added some thinger2cisml.pl parsers
	

=cut