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

# Read hits
$ok = 0;
while ($line = <>) {
    chomp $line;
    if ($line =~ /sequence file:\s+(.*)/) { $sequence_file = $1; }
    elsif ($line =~ /^-+$/) { $ok = 1; }
    elsif ($ok) {
        if ($line =~ /Sequence:\s+(.*)/) { $seq_name = $1; }
        elsif ($line =~ /^\s*$/) { last; }
        else {
	($pattern_acc, $hit_loc, $core_sim, $mat_sim, $hit_seq) = split(/\s+\|\s+/,$line);
	$pattern_acc =~ s/\s//g;
	($loc_number,$dir) = $hit_loc =~ /(\d+)\s+\(([+-])\)/;
	if ($dir eq "+") {
	    $hit_start = $loc_number;
	    $hit_stop = ($loc_number + length($hit_seq)) - 1;
	}
	else {
	    $hit_start = $loc_number;
	    $hit_stop = ($loc_number - length($hit_seq)) + 1; 
	}
	unless (defined($patterns{$pattern_acc})) { $patterns{$pattern_acc} = $pattern_acc; }
	push(@{$hits{$pattern_acc}}, [$hit_start, $hit_stop, $hit_seq, $core_sim, $mat_sim, $seq_name]);
        }
    }
}

#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 "xmlns:mi=\"MatInspector\"\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>MatInspector</program-name>\n";
print "\t<parameters>\n\t\t<sequence-file>$sequence_file</sequence-file>\n\t</parameters>\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\">\n";
    foreach $match (@{$hits{$pattern}}) {
        print "\t\t\t<matched-element start=\"$$match[0]\" stop=\"$$match[1]\" score=\"$$match[4]\">\n";
        print "\t\t\t\t<sequence>$$match[2]</sequence>\n";
        print "\t\t\t\t<mi:cor-sim>$$match[3]</mi:cor-sim>\n";
        print "\t\t\t</matched-element>\n";
    }
    print "\t\t</scanned-sequence>\n";
    print "\t</pattern>\n";
}

#Print CisML footer
print "</cis-element-search>\n";
=head1 NAME

tfscan2cisml.pl

=head1 SYNOPSIS

tfscan2cisml.pl

=head1 DESCRIPTION

Converts output of MatInspector (http://www.gsf.de/biodv/matinspector.html) to CisML XML format.
MatInspector output does not provide all of the information generally included in CisML.

matinspector can be obtained from 

matinspector output looks like this:

MatInspector Release 1.0   September 1995           Thu Nov 16 11:48:07 1995


   Solution parameters:
   ~~~~~~~~~~~~~~~~~~~~
   sequence file:     chr3.seq
   selected matrices: 
            F$ABF1_01            (ABF1                          )  re:   0.13
            	core sim:   0.80     matrix sim: 0.85

   Explanation for column output:
   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	->	Matrix positions correspond to sense strand numbering, but all
		sequences are given in 5'-3' direction.

	->	n/a in column 'core simil.' indicates, that no core search
		was conducted.

	->	n/a in column 'matrix simil.' indicates, that no 
		matrix similarity was calculated, because the sequences was
		too short (only the core was found). In that case the
		core position is given in brackets in column 'matrix position'.

	->	Capital letters within the sequence indicate the core string.

 matrix               | matrix       | core   | matrix | sequence
 name                 | position(str)| simil. | simil. | 
-----------------------------------------------------------------------
 Sequence: SCCHRIII
 F$ABF1_01            |      820 (+) |  0.951 |  0.930 | tcgcatcattatgcACGGcttg
 F$ABF1_01            |     8203 (+) |  0.976 |  0.876 | cgatgtcatagagtACGTgtca
 F$ABF1_01            |    14114 (-) |  1.000 |  0.914 | acacatccttaaatACGAaagt
 F$ABF1_01            |    14584 (-) |  0.976 |  0.931 | ggatatcattgcaaACGTcggg
 F$ABF1_01            |    17119 (-) |  0.951 |  0.877 | actaatcaccgcgaACGGaaac
 F$ABF1_01            |    24619 (+) |  1.000 |  0.857 | tagtatcgcgctgcACGAgcgt

...

 F$ABF1_01            |   293524 (-) |  1.000 |  0.908 | ttatatcgccatatACGAaaat
 F$ABF1_01            |   301286 (+) |  0.976 |  0.861 | ttgggtcatataaaACGTctgc
 F$ABF1_01            |   304162 (-) |  0.951 |  0.901 | tgtcgtcattgaagACGGtaaa
 F$ABF1_01            |   309562 (+) |  1.000 |  0.925 | acaagtcattgagaACGAaatt
 F$ABF1_01            |   314756 (-) |  0.951 |  0.959 | ccacatcattatgcACGGcact

 In   1 sequences  85 matches to the matrix F$ABF1_01 were found.

=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: matinspector2cisml.pl,v 1.4 2004/02/22 21:41:37 phaverty Exp $

	$Log: matinspector2cisml.pl,v $
	Revision 1.4  2004/02/22 21:41:37  phaverty
	dunno
	
	Revision 1.3  2004/02/18 15:26:56  phaverty
	Changed sequence in matched-element to a sequence element.
	
	Revision 1.2  2004/02/18 02:50:14  phaverty
	changed documentation
	
	Revision 1.1  2004/02/18 02:41:27  phaverty
	Added some thinger2cisml.pl parsers
	

=cut