libbio-perl-perl 1.6.924-1 / usr / share / perl5 / Bio / SeqIO / bsml.pm

#
# BioPerl module for Bio::SeqIO::bsml
#
# Please direct questions and support issues to <bioperl-l@bioperl.org> 
#
# Cared for by Charles Tilford (tilfordc@bms.com)
# Copyright (C) Charles Tilford 2001
#
# This library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License as published by the Free Software Foundation; either
# version 2.1 of the License, or (at your option) any later version.
#
# This library is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
# Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser General Public
# License along with this library; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
# Also at:   http://www.gnu.org/copyleft/lesser.html


# Much of the basic documentation in this module has been
# cut-and-pasted from the embl.pm (Ewan Birney) SeqIO module.


=head1 NAME

Bio::SeqIO::bsml - BSML sequence input/output stream

=head1 SYNOPSIS

 It is probably best not to use this object directly, but rather go
 through the SeqIO handler system. To read a BSML file:

    $stream = Bio::SeqIO->new( -file => $filename, -format => 'bsml');

    while ( my $bioSeqObj = $stream->next_seq() ) {
   	# do something with $bioSeqObj
    }

 To write a Seq object to the current file handle in BSML XML format:

    $stream->write_seq( -seq => $seqObj);

 If instead you would like a XML::DOM object containing the BSML, use:

    my $newXmlObject = $stream->to_bsml( -seq => $seqObj);

=head1 DEPENDENCIES

 In addition to parts of the Bio:: hierarchy, this module uses:

 XML::DOM

=head1 DESCRIPTION

 This object can transform Bio::Seq objects to and from BSML (XML)
 flatfiles.

=head2 NOTE:

 2/1/02 - I have changed the API to more closely match argument
 passing used by other BioPerl methods ( -tag => value ). Internal
 methods are using the same API, but you should not be calling those
 anyway...

=head1 FEEDBACK

=head2 Mailing Lists

User feedback is an integral part of the evolution of this and other
Bioperl modules. Send your comments and suggestions preferably to one
of the Bioperl mailing lists.  Your participation is much appreciated.

  bioperl-l@bioperl.org                  - General discussion
  http://bioperl.org/wiki/Mailing_lists  - About the mailing lists

=head2 Support 

Please direct usage questions or support issues to the mailing list:

I<bioperl-l@bioperl.org>

rather than to the module maintainer directly. Many experienced and 
reponsive experts will be able look at the problem and quickly 
address it. Please include a thorough description of the problem 
with code and data examples if at all possible.

=head2 Reporting Bugs

 Report bugs to the Bioperl bug tracking system to help us keep track
 the bugs and their resolution.
 Bug reports can be submitted via the web:

  https://github.com/bioperl/bioperl-live/issues

=head2 Things Still to Do

 * The module now uses the new Collection.pm system. However,
   Annotations associated with a Feature object still seem to use the
   old system, so parsing with the old methods are included..

 * Generate Seq objects with no sequence data but an assigned
   length. This appears to be an issue with Bio::Seq. It is possible
   (and reasonable) to make a BSML document with features but no
   sequence data.

 * Support <Seq-data-import>. Do not know how commonly this is used.

 * Some features are awaiting implementation in later versions of
   BSML. These include:

       * Nested feature support

       * Complex feature (ie joins)

       * Unambiguity in strand (ie -1,0,1, not just  'complement' )

       * More friendly dblink structures

 * Location.pm (or RangeI::union?) appears to have a bug when 'expand'
   is used.

 * More intelligent hunting for sequence and feature titles? It is not
   terribly clear where the most appropriate field is located, better
   grepping (eg looking for a reasonable count for spaces and numbers)
   may allow for titles better than "AE008041".

=head1 AUTHOR - Charles Tilford

Bristol-Myers Squibb Bioinformatics

Email tilfordc@bms.com

I have developed the BSML specific code for this package, but have used
code from other SeqIO packages for much of the nuts-and-bolts. In particular
I have used code from the embl.pm module either directly or as a framework
for many of the subroutines that are common to SeqIO modules.

=cut

package Bio::SeqIO::bsml;
use strict;

use Bio::SeqFeature::Generic;
use Bio::Species;
use XML::DOM;
use Bio::Seq::SeqFactory;
use Bio::Annotation::Collection;
use Bio::Annotation::Comment;
use Bio::Annotation::Reference;
use Bio::Annotation::DBLink;

use base qw(Bio::SeqIO);

my $idcounter = {};  # Used to generate unique id values
my $nvtoken = ": ";  # The token used if a name/value pair has to be stuffed
                     # into a single line

=head1 METHODS

=cut

# LS: this seems to get overwritten on line 1317, generating a redefinition error.  Dead code?
# CAT: This was inappropriately added in revision 1.10 - I added the check for existance of a sequence factory to the actual _initialize
# sub _initialize {
#   my($self,@args) = @_;
#   $self->SUPER::_initialize(@args);
#   if( ! defined $self->sequence_factory ) {
#       $self->sequence_factory(Bio::Seq::SeqFactory->new(-verbose => $self->verbose(), -type => 'Bio::Seq::RichSeq'));
#   }
# }

=head2 next_seq

 Title   : next_seq
 Usage   : my $bioSeqObj = $stream->next_seq
 Function: Retrieves the next sequence from a SeqIO::bsml stream.
 Returns : A reference to a Bio::Seq::RichSeq object
 Args    :

=cut

sub next_seq {
    my $self = shift;
    my ($desc);
    my $bioSeq = $self->sequence_factory->create(-verbose =>$self->verbose());

    unless (exists $self->{'domtree'}) {
	$self->throw("A BSML document has not yet been parsed.");
	return;
    }
    my $dom = $self->{'domtree'};
    my $seqElements = $dom->getElementsByTagName ("Sequence");
    if ($self->{'current_node'} == $seqElements->getLength ) {
	# There are no more <Sequence>s to process
	return;
    }
    my $xmlSeq = $seqElements->item($self->{'current_node'});

    # Assume that title attribute contains the best display id
    if (my $val = $xmlSeq->getAttribute( "title")) {
	$bioSeq->display_id($val);
    }

    # Set the molecule type
    if (my $val = $xmlSeq->getAttribute( "molecule" )) {
	my %mol = ('dna' => 'DNA', 'rna' => 'RNA', 'aa' => 'protein');
	$bioSeq->molecule($mol{ lc($val) });
    }

    # Set the accession number
    if (my $val = $xmlSeq->getAttribute( "ic-acckey" )) {
	$bioSeq->accession_number($val);
    }

    # Get the sequence data for the element
    if (my $seqData = &FIRSTDATA($xmlSeq->getElementsByTagName("Seq-data")
				 ->item(0) ) ) {
	# Sequence data exists, transfer to the Seq object
	# Remove white space and CRs (not neccesary?)
	$seqData =~ s/[\s\n\r]//g;
	$bioSeq->seq($seqData);
    } elsif (my $import = $xmlSeq->getElementsByTagName("Seq-dataimport")
	     ->item(0) )  {
#>>>>  # What about <Seq-data-import> ??

    } elsif (my $val = $xmlSeq->getAttribute("length"))  {
	# No sequence defined, set the length directly

#>>>>   # This does not appear to work - length is apparently calculated
	# from the sequence. How to make a "virtual" sequence??? Such
	# creatures are common in BSML...
	$bioSeq->length($val);
    }

    my $species = Bio::Species->new();
    my @classification = ();

    # Peruse the generic <Attributes> - those that are direct children of
    # the <Sequence> or the <Feature-tables> element
    # Sticky wicket here - data not controlled by schema, could be anything
    my @seqDesc = ();
    my %specs = ('common_name' => 'y',
		 'genus'       => 'y',
		 'species'     => 'y',
		 'sub_species' => 'y', 
		 );
    my %seqMap = (
		  'add_date'     => [ qw(date date-created date-last-updated)],
		  'keywords'     => [ 'keyword', ],
		  'seq_version'  => [ 'version' ],
		  'division'     => [ 'division' ],
		  'add_secondary_accession' => ['accession'],
		  'pid'          => ['pid'],
		  'primary_id'   => [ 'primary.id', 'primary_id' ],
		  );
    my @links;
    my $floppies = &GETFLOPPIES($xmlSeq);
    for my $attr (@{$floppies}) {
	# Don't want to get attributes from <Feature> or <Table> elements yet
	my $parent = $attr->getParentNode->getNodeName;
	next unless($parent eq "Sequence" || $parent eq "Feature-tables");

	my ($name, $content) = &FLOPPYVALS($attr);
	$name = lc($name);
	if (exists $specs{$name}) { # It looks like part of species...
	    $species->$name($content);
	    next;
	}
	my $value = "";
	# Cycle through the Seq methods:
	for my $method (keys %seqMap) {
	    # Cycle through potential matching attributes:
	    for my $match (@{$seqMap{$method}}) {
		# If the <Attribute> name matches one of the keys,
		# set $value, unless it has already been set
		$value ||= $content if ($name =~ /$match/i);
	    }
	    if ($value ne "") {
		
		if( $method eq 'seq_version'&& $value =~ /\S+\.(\d+)/ ) {
		    # hack for the fact that data in version is actually
		    # ACCESSION.VERSION
		    ($value) = $1;
		}
		$bioSeq->$method($value);
		last;
	    }
	}
	if( $name eq 'database-xref' ) {
	    my ($link_id,$link_db) = split(/:/,$value);
	    push @links, Bio::Annotation::DBLink->new(-primary_id => $link_id,
						      -database   => $link_db);
	}
	next if ($value ne "");

	if ($name =~ /^species$/i) { # Uh, it's the species designation?
	    if ($content =~ / /) {
		# Assume that a full species name has been provided
		# This will screw up if the last word is the subspecies...
		my @break = split " ", $content;
		@classification = reverse @break;
	    } else {
		$classification[0] = $content;
	    }
	    next;
	}
	if ($name =~ /sub[_ ]?species/i) { # Should be the subspecies...
	    $species->sub_species( $content );
	    next;
	}
	if ($name =~ /classification/i) { # Should be species classification
	    # We will assume that there are spaces separating the terms:
	    my @bits = split " ", $content;
	    # Now make sure there is not other cruft as well (eg semi-colons)
	    for my $i (0..$#bits) {
		$bits[$i] =~ /(\w+)/;
		$bits[$i] = $1;
	    }
	    $species->classification( @bits );
	    next;
	}
	if ($name =~ /comment/) {
	    my $com = Bio::Annotation::Comment->new('-text' => $content);
	    #  $bioSeq->annotation->add_Comment($com);
	    $bioSeq->annotation->add_Annotation('comment', $com);
	    next;
	}
	# Description line - collect all descriptions for later assembly
	if ($name =~ /descr/) {
	    push @seqDesc, $content;
	    next;
	}
	# Ok, we have no idea what this attribute is. Dump to SimpleValue
	my $simp = Bio::Annotation::SimpleValue->new( -value => $content);
	$bioSeq->annotation->add_Annotation($name, $simp);
    }
    unless ($#seqDesc < 0) {
	$bioSeq->desc( join "; ", @seqDesc);
    }

#>>>>  This should be modified so that any IDREF associated with the
    # <Reference> is then used to associate the reference with the
    # appropriate Feature

    # Extract out <Reference>s associated with the sequence
    my @refs;
    my %tags = (
		-title    => "RefTitle",
		-authors  => "RefAuthors",
		-location => "RefJournal",
		);
    for my $ref ( $xmlSeq->getElementsByTagName ("Reference") ) {
	my %refVals;
	for my $tag (keys %tags) {
	    my $rt = &FIRSTDATA($ref->getElementsByTagName($tags{$tag})
				->item(0));
	    next unless ($rt);
	    $rt =~ s/^[\s\r\n]+//; # Kill leading space
	    $rt =~ s/[\s\r\n]+$//; # Kill trailing space
	    $rt =~ s/[\s\r\n]+/ /; # Collapse internal space runs
	    $refVals{$tag} = $rt;
	}
	my $reference = Bio::Annotation::Reference->new( %refVals );
	
	# Pull out any <Reference> information hidden in <Attributes>
	my %refMap = (
		      comment         => [ 'comment', 'remark' ],
		      medline         => [ 'medline', ],
		      pubmed          => [ 'pubmed' ],
		      start           => [ 'start', 'begin' ],
		      end             => [ 'stop', 'end' ],		      
		      );
	my @refCom = ();
	my $floppies = &GETFLOPPIES($ref);
	for my $attr (@{$floppies}) {
	    my ($name, $content) = &FLOPPYVALS($attr);	    
	    my $value = "";
	    # Cycle through the Seq methods:
	    for my $method (keys %refMap) {
		# Cycle through potential matching attributes:
		for my $match (@{$refMap{$method}}) {
		    # If the <Attribute> name matches one of the keys,
		    # set $value, unless it has already been set
		    $value ||= $content if ($name =~ /$match/i);
		}
		if ($value ne "") {
		    my $str = '$reference->' . $method . "($value)";
		    eval($str);
		    next;
		}
	    }
	    next if ($value ne "");
	    # Don't know what the <Attribute> is, dump it to comments:
	    push @refCom, $name . $nvtoken . $content;
	}
	unless ($#refCom < 0) {
	    # Random stuff was found, tack it to the comment field
	    my $exist = $reference->comment;
	    $exist .= join ", ", @refCom;
	    $reference->comment($exist);
	}
	push @refs, $reference;
    }
    $bioSeq->annotation->add_Annotation('reference' => $_) for @refs;
    my $ann_col = $bioSeq->annotation;
    # Extract the <Feature>s for this <Sequence>
    for my $feat ( $xmlSeq->getElementsByTagName("Feature") ) {
	$bioSeq->add_SeqFeature( $self->_parse_bsml_feature($feat) );
    }
    
    $species->classification( @classification );
    $bioSeq->species( $species );
    
    $bioSeq->annotation->add_Annotation('dblink' => $_) for @links;
    
    $self->{'current_node'}++;
    return $bioSeq;
}
#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Get all the <Attribute> and <Qualifier> children for an object, and
# return them as an array reference
# ('floppy' since these elements have poor/no schema control)
sub GETFLOPPIES {
    my $obj = shift;

    my @floppies;
    my $attributes = $obj->getElementsByTagName ("Attribute");
    for (my $i = 0; $i < $attributes->getLength; $i++) {
	push @floppies, $attributes->item($i);
    }
    my $qualifiers = $obj->getElementsByTagName ("Qualifier");
    for (my $i = 0; $i < $qualifiers->getLength; $i++) {
	push @floppies, $qualifiers->item($i);
    }
    return \@floppies;
}
#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Given a DOM <Attribute> or <Qualifier> object, return the [name, value] pair
sub FLOPPYVALS {
    my $obj = shift;

    my ($name, $value);
    if      ($obj->getNodeName eq "Attribute") {
	$name  = $obj->getAttribute('name');
	$value = $obj->getAttribute('content');
    } elsif ($obj->getNodeName eq "Qualifier") {
	# Wheras <Attribute>s require both 'name' and 'content' attributes,
	# <Qualifier>s can technically have either blank (and sometimes do)
	my $n =  $obj->getAttribute('value-type');
	$name = $n if ($n ne "");
	my $v =  $obj->getAttribute('value');
	$value = $v if ($v ne "");
    }
    return ($name, $value);
}
#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Returns the value of the first TEXT_NODE encountered below an element
# Rational - avoid grabbing a comment rather than the PCDATA. Not foolproof...
sub FIRSTDATA {
    my $element = shift;
    return unless ($element);

    my $hopefuls = $element->getChildNodes;
    my $data;
    for (my $i = 0; $i < $hopefuls->getLength; $i++) {
	if ($hopefuls->item($i)->getNodeType ==
	  XML::DOM::Node::TEXT_NODE() ) {
	    $data = $hopefuls->item($i)->getNodeValue;
	    last;
	}
    }
    return $data;
}
#- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Just collapses whitespace runs in a string
sub STRIP {
    my $string = shift;
    $string =~ s/[\s\r\n]+/ /g;
    return $string;
}

=head2 to_bsml

 Title   : to_bsml
 Usage   : my $domDoc = $obj->to_bsml(@args)
 Function: Generates an XML structure for one or more Bio::Seq objects.
           If $seqref is an array ref, the XML tree generated will include
           all the sequences in the array.
 Returns : A reference to the XML DOM::Document object generated / modified
 Args    : Argument array in form of -key => val. Recognized keys:

      -seq A Bio::Seq reference, or an array reference of many of them

   -xmldoc Specifies an existing XML DOM document to add the sequences
           to. If included, then only data (no page formatting) will
           be added. If not, a new XML::DOM::Document will be made,
           and will be populated with both <Sequence> data, as well as
           <Page> display elements.

   -nodisp Do not generate <Display> elements, or any children
           thereof, even if -xmldoc is not set.

 -skipfeat If set to 'all', all <Feature>s will be skipped.  If it is
           a hash reference, any <Feature> with a class matching a key
           in the hash will be skipped - for example, to skip 'source'
           and 'score' features, use:

               -skipfeat => { source => 'Y', score => 'Y' }

 -skiptags As above: if set to 'all', no tags are included, and if a
           hash reference, those specific tags will be ignored.

           Skipping some or all tags and features can result in
           noticeable speed improvements.

   -nodata If true, then <Seq-data> will not be included.  This may be
           useful if you just want annotations and do not care about
           the raw ACTG information.

   -return Default is 'xml', which will return a reference to the BSML
           XML object. If set to 'seq' will return an array ref of the
           <Sequence> objects added (rather than the whole XML object)

    -close Early BSML browsers will crash if an element *could* have
           children but does not, and is closed as an empty element
           e.g. <Styles/>. If -close is true, then such tags are given
           a comment child to explicitly close them e.g.  <Styles><!--
           --></Styles>. This is default true, set to "0" if you do
           not want this behavior.

 Examples : my $domObj = $stream->to_bsml( -seq => \@fourCoolSequenceObjects,
					   -skipfeat => { source => 1 },
					   );

            # Or add sequences to an existing BSML document:
            $stream->to_bsml( -seq => \@fourCoolSequenceObjects,
			      -skipfeat => { source => 1 },
			      -xmldoc => $myBsmlDocumentInProgress,  );

=cut

sub to_bsml {
    my $self = shift;
    my $args = $self->_parseparams( -close => 1,
				    -return => 'xml',
				    @_);
    $args->{NODISP} ||= $args->{NODISPLAY};
    my $seqref = $args->{SEQ};
    $seqref = (ref($seqref) eq 'ARRAY') ? $seqref : [ $seqref ];

    #############################
    # Basic BSML XML Components #
    #############################

    my $xml;
    my ($bsmlElem, $defsElem, $seqsElem, $dispElem);
    if ($args->{XMLDOC}) {
	# The user has provided an existing XML DOM object
	$xml = $args->{XMLDOC};
	unless ($xml->isa("XML::DOM::Document")) {
	    $self->throw('SeqIO::bsml.pm error:\n'.
		 'When calling ->to_bsml( { xmldoc => $myDoc }), $myDoc \n' .
		 'should be an XML::DOM::Document object, or an object that\n'.
		 'inherits from that class (like BsmlHelper.pm)');
	}
    } else {
	# The user has not provided a new document, make one from scratch
	$xml = XML::DOM::Document->new();
	$xml->setXMLDecl( $xml->createXMLDecl("1.0") );
	my $url = "http://www.labbook.com/dtd/bsml2_2.dtd";
	my $doc = $xml->createDocumentType("Bsml",$url);
	$xml->setDoctype($doc);
	$bsmlElem = $self->_addel( $xml, 'Bsml');
	$defsElem = $self->_addel( $bsmlElem, 'Definitions');
	$seqsElem = $self->_addel( $defsElem, 'Sequences');
	unless ($args->{NODISP}) {
	    $dispElem = $self->_addel( $bsmlElem, 'Display');
	    my $stylElem = $self->_addel( $dispElem, 'Styles');
	    my $style = $self->_addel( $stylElem, 'Style', {
		type => "text/css" });
	    my $styleText =
		qq(Interval-widget { display : "1"; }\n) .
		    qq(Feature { display-auto : "1"; });
	    $style->appendChild( $xml->createTextNode($styleText) );
	}
    }

    # Establish fundamental BSML elements, if they do not already exist
    $bsmlElem ||= $xml->getElementsByTagName("Bsml")->item(0);
    $defsElem ||= $xml->getElementsByTagName("Definitions")->item(0);
    $seqsElem ||= $xml->getElementsByTagName("Sequences")->item(0);

    ###############
    # <Sequences> #
    ###############

    # Map over Bio::Seq to BSML
    my %mol = ('dna' => 'DNA', 'rna' => 'RNA', 'protein' => 'AA');
    my @xmlSequences;

    for my $bioSeq (@{$seqref}) {
	my $xmlSeq = $xml->createElement("Sequence");
	my $FTs    = $xml->createElement("Feature-tables");

	# Array references to hold <Reference> objects:
	my $seqRefs = []; my $featRefs = [];
	# Array references to hold <Attribute> values (not objects):
	my $seqDesc = [];
	push @{$seqDesc}, ["comment" , "This file generated to BSML 2.2 standards - joins will be collapsed to a single feature enclosing all members of the join"];
	push @{$seqDesc}, ["description" , eval{$bioSeq->desc}];
	for my $kwd ( eval{$bioSeq->get_keywords} ) {
	    push @{$seqDesc}, ["keyword" , $kwd];
	}
	push @{$seqDesc}, ["keyword" , eval{$bioSeq->keywords}];
	push @{$seqDesc}, ["version" , eval{
	    join(".", $bioSeq->accession_number, $bioSeq->seq_version); }];
	push @{$seqDesc}, ["division" , eval{$bioSeq->division}];
	push @{$seqDesc}, ["pid" , eval{$bioSeq->pid}];
#	push @{$seqDesc}, ["bio_object" , ref($bioSeq)];
	push @{$seqDesc}, ["primary_id" , eval{$bioSeq->primary_id}];
	for my $dt (eval{$bioSeq->get_dates()} ) {
	    push @{$seqDesc}, ["date" , $dt];
	}
	for my $ac (eval{$bioSeq->get_secondary_accessions()} ) {
	    push @{$seqDesc}, ["secondary_accession" , $ac];
	}

	# Determine the accession number and a unique identifier
	my $acc = $bioSeq->accession_number eq "unknown" ?
	    "" : $bioSeq->accession_number;
	my $id;
	my $pi = $bioSeq->primary_id;
	if ($pi && $pi !~ /Bio::/) {
	    # Not sure I understand what primary_id is... It sometimes
	    # is a string describing a reference to a BioSeq object...
	    $id = "SEQ" . $bioSeq->primary_id;
	} else {
	    # Nothing useful found, make a new unique ID
	    $id = $acc || ("SEQ-io" . $idcounter->{Sequence}++);
	}
	# print "$id->",ref($bioSeq->primary_id),"\n";
	# An id field with spaces is interpreted as an idref - kill the spaces
	$id =~ s/ /-/g;
	# Map over <Sequence> attributes
	my %attr = ( 'title'         => $bioSeq->display_id,
		     'length'        => $bioSeq->length,
		     'ic-acckey'     => $acc,
		     'id'            => $id,
		     'representation' => 'raw',
		     );
	$attr{molecule} = $mol{ lc($bioSeq->molecule) } if $bioSeq->can('molecule');


	for my $a (keys %attr) {
	    $xmlSeq->setAttribute($a, $attr{$a}) if (defined $attr{$a} &&
						     $attr{$a} ne "");
	}
	# Orphaned Attributes:
	$xmlSeq->setAttribute('topology', 'circular')
	    if ($bioSeq->is_circular);
	# <Sequence> strand, locus

	$self->_add_page($xml, $xmlSeq) if ($dispElem);
	################
	# <Attributes> #
	################

	# Check for Bio::Annotations on the * <Sequence> *.
	$self->_parse_annotation( -xml => $xml, -obj => $bioSeq,
				  -desc => $seqDesc, -refs => $seqRefs);

	# Incorporate species data
	if (ref($bioSeq->species) eq 'Bio::Species') {
	    # Need to peer into Bio::Species ...
	    my @specs = ('common_name', 'genus', 'species', 'sub_species');
	    for my $sp (@specs) {
		next unless (my $val = $bioSeq->species()->$sp());
		push @{$seqDesc}, [$sp , $val];
	    }
	    push @{$seqDesc}, ['classification',
			       (join " ", $bioSeq->species->classification) ];
	    # Species::binomial will return "genus species sub_species" ...
	} elsif (my $val = $bioSeq->species) {
	    # Ok, no idea what it is, just dump it in there...
	    push @{$seqDesc}, ["species", $val];
	}

	# Add the description <Attribute>s for the <Sequence>
	for my $seqD (@{$seqDesc}) {
	    $self->_addel($xmlSeq, "Attribute", {
		name => $seqD->[0], content => $seqD->[1]}) if ($seqD->[1]);
	}

	# If sequence references were added, make a Feature-table for them
	unless ($#{$seqRefs} < 0) {
	    my $seqFT = $self->_addel($FTs, "Feature-table", {
		title => "Sequence References", });
	    for my $feat (@{$seqRefs}) {
		$seqFT->appendChild($feat);
	    }
	}

	# This is the appropriate place to add <Feature-tables>
	$xmlSeq->appendChild($FTs);

	#############
	# <Feature> #
	#############

#>>>>	# Perhaps it is better to loop through top_Seqfeatures?...
#>>>>	# ...however, BSML does not have a hierarchy for Features

	if (defined $args->{SKIPFEAT} &&
	    $args->{SKIPFEAT} eq 'all') {
	    $args->{SKIPFEAT} = { all => 1};
	} else { $args->{SKIPFEAT} ||= {} }
	for my $class (keys %{$args->{SKIPFEAT}}) {
	    $args->{SKIPFEAT}{lc($class)} = $args->{SKIPFEAT}{$class};
	}
	# Loop through all the features
	my @features = $bioSeq->all_SeqFeatures();
	if (@features && !$args->{SKIPFEAT}{all}) {
	    my $ft = $self->_addel($FTs, "Feature-table", {
		title => "Features", });
	    for my $bioFeat (@features ) {
		my $featDesc = [];
		my $class = lc($bioFeat->primary_tag);
		# The user may have specified to ignore this type of feature
		next if ($args->{SKIPFEAT}{$class});
		my $id = "FEAT-io" . $idcounter->{Feature}++;
		my $xmlFeat = $self->_addel( $ft, 'Feature', {
		    'id' => $id,
		    'class' => $class ,
		    'value-type' => $bioFeat->source_tag });
		# Check for Bio::Annotations on the * <Feature> *.
		$self->_parse_annotation( -xml => $xml, -obj => $bioFeat,
					  -desc => $featDesc, -id => $id,
					  -refs =>$featRefs, );
		# Add the description stuff for the <Feature>
		for my $de (@{$featDesc}) {
		    $self->_addel($xmlFeat, "Attribute", {
			name => $de->[0], content => $de->[1]}) if ($de->[1]);
		}
		$self->_parse_location($xml, $xmlFeat, $bioFeat);

		# loop through the tags, add them as <Qualifiers>
		next if (defined $args->{SKIPTAGS} &&
			 $args->{SKIPTAGS} =~ /all/i);
		# Tags can consume a lot of CPU cycles, and can often be
		# rather non-informative, so -skiptags can allow total or
		# selective omission of tags.
		for my $tag ($bioFeat->all_tags()) {
		    next if (exists $args->{SKIPTAGS}{$tag});
		    for my $val ($bioFeat->each_tag_value($tag)) {
			$self->_addel( $xmlFeat, 'Qualifier', {
			    'value-type' => $tag ,
			    'value' => $val });
		    }
		}
	    }
	}

	##############
	# <Seq-data> #
	##############

	# Add sequence data
	if ( (my $data = $bioSeq->seq) && !$args->{NODATA} ) {
	    my $d = $self->_addel($xmlSeq, 'Seq-data');
	    $d->appendChild( $xml->createTextNode($data) );
	}

	# If references were added, make a Feature-table for them
	unless ($#{$featRefs} < 0) {
	    my $seqFT = $self->_addel($FTs, "Feature-table", {
		title => "Feature References", });
	    for my $feat (@{$featRefs}) {
		$seqFT->appendChild($feat);
	    }
	}

	# Place the completed <Sequence> tree as a child of <Sequences>
	$seqsElem->appendChild($xmlSeq);
	push @xmlSequences, $xmlSeq;
    }

    # Prevent browser crashes by explicitly closing empty elements:
    if ($args->{CLOSE}) {
	my @problemChild = ('Sequences', 'Sequence', 'Feature-tables',
			    'Feature-table', 'Screen', 'View',);
	for my $kid (@problemChild) {
	    for my $prob ($xml->getElementsByTagName($kid)) {
		unless ($prob->hasChildNodes) {
		    $prob->appendChild(
			$xml->createComment(" Must close <$kid> explicitly "));
		}
	    }
	}
    }

    if (defined $args->{RETURN} &&
	$args->{RETURN} =~ /seq/i) {
	return \@xmlSequences;
    } else {
	return $xml;
    }
}

=head2 write_seq

 Title   : write_seq
 Usage   : $obj->write_seq(@args)
 Function: Prints out an XML structure for one or more Bio::Seq objects.
           If $seqref is an array ref, the XML tree generated will include
           all the sequences in the array. This method is fairly simple,
           most of the processing is performed within to_bsml.
 Returns : A reference to the XML object generated / modified
 Args    : Argument array. Recognized keys:

      -seq A Bio::Seq reference, or an array reference of many of them

           Alternatively, the method may be called simply as...

           $obj->write_seq( $bioseq )

           ... if only a single argument is passed, it is assumed that
           it is the sequence object (can also be an array ref of
           many Seq objects )

-printmime If true prints "Content-type: $mimetype\n\n" at top of
           document, where $mimetype is the value designated by this
           key. For generic XML use text/xml, for BSML use text/x-bsml

   -return This option will be supressed, since the nature of this
           method is to print out the XML document. If you wish to
           retrieve the <Sequence> objects generated, use the to_bsml
           method directly.

=cut

sub write_seq {
    my $self = shift;
    my $args = $self->_parseparams( @_);
    if ($#_ == 0 ) {
	# If only a single value is passed, assume it is the seq object
	unshift @_, "-seq";
    }
    # Build a BSML XML DOM object based on the sequence(s)
    my $xml = $self->to_bsml( @_,
			      -return => undef );
    # Convert to a string
    my $out = $xml->toString;
    # Print after putting a return after each element - more readable
    $out =~ s/>/>\n/g;
    $self->_print("Content-type: " . $args->{PRINTMIME} . "\n\n")
	if ($args->{PRINTMIME});
    $self->_print( $out );
    # Return the DOM tree in case the user wants to do something with it

    $self->flush if $self->_flush_on_write && defined $self->_fh;
    return $xml;
}

=head1 INTERNAL METHODS
#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-

 The following methods are used for internal processing, and should probably
 not be accessed by the user.

=head2 _parse_location

 Title   : _parse_location
 Usage   : $obj->_parse_location($xmlDocument, $parentElem, $SeqFeatureObj)
 Function: Adds <Interval-loc> and <Site-loc> children to <$parentElem> based
           on locations / sublocations found in $SeqFeatureObj. If
           sublocations exist, the original location will be ignored.
 Returns : An array ref containing the elements added to the parent.
           These will have already been added to <$parentElem>
 Args    : 0 The DOM::Document being modified
           1 The DOM::Element parent that you want to add to
           2 Reference to the Bio::SeqFeature being analyzed

=cut

    ###############################
    # <Interval-loc> & <Site-loc> #
    ###############################

sub _parse_location {
    my $self = shift;
    my ($xml, $xmlFeat, $bioFeat) = @_;
    my $bioLoc = $bioFeat->location;
    my @locations;
    if (ref($bioLoc) =~ /Split/) {
	@locations = $bioLoc->sub_Location;
	# BSML 2.2 does not recognize / support joins. For this reason,
	# we will just use the upper-level location. The line below can
	# be deleted or commented out if/when BSML 3 supports complex
	# interval deffinitions:
	@locations = ($bioLoc);
    } else {
	@locations = ($bioLoc);
    }
    my @added = ();

    # Add the site or interval positional information:
    for my $loc (@locations) {
	my ($start, $end) = ($loc->start, $loc->end);
	my %locAttr;
	# Strand information is not well described in BSML
	$locAttr{complement} = 1 if ($loc->strand == -1);
	if ($start ne "" && ($start == $end || $end eq "")) {
	    $locAttr{sitepos} = $start;
	    push @added, $self->_addel($xmlFeat,'Site-loc',\%locAttr);
	} elsif ($start ne "" && $end ne "") {
	    if ($start > $end) {
		# The feature is on the complementary strand
		($start, $end) = ($end, $start);
		$locAttr{complement} = 1;
	    }
	    $locAttr{startpos} = $start;
	    $locAttr{endpos} = $end;
	    push @added, $self->_addel($xmlFeat,'Interval-loc',\%locAttr);
	} else {
	    warn "Failure to parse SeqFeature location. Start = '$start' & End = '$end'";
	}
    }
    return \@added;
}

=head2 _parse_bsml_feature

 Title   : _parse_bsml_feature
 Usage   : $obj->_parse_bsml_feature($xmlFeature )
 Function: Will examine the <Feature> element provided by $xmlFeature and
           return a generic seq feature.
 Returns : Bio::SeqFeature::Generic
 Args    : 0 XML::DOM::Element <Feature> being analyzed.

=cut

sub _parse_bsml_feature {
    my $self = shift;
    my ($feat) = @_;

    my $basegsf = Bio::SeqFeature::Generic->new();
       # score
       # frame
       # source_tag

    # Use the class as the primary tag value, if it is present
    if ( my $val = $feat->getAttribute("class") ) {
	$basegsf->primary_tag($val);
    }

    # Positional information is in <Interval-loc>s or <Site-loc>s
    # We need to grab these in order, to try to recreate joins...
    my @locations = ();
    for my $kid ($feat->getChildNodes) {
	my $nodeName = $kid->getNodeName;
	next unless ($nodeName eq "Interval-loc" ||
		     $nodeName eq "Site-loc");
	push @locations, $kid;
    }
    if ($#locations == 0) {
	# There is only one location specified
	$self->_parse_bsml_location($locations[0], $basegsf);
    } elsif ($#locations > 0) {
#>>>>   # This is not working, I think the error is somewhere downstream
	# of add_sub_SeqFeature, probably in RangeI::union ?
	# The sub features are added fine, but the EXPANDed parent feature
	# location has a messed up start - Bio::SeqFeature::Generic ref
	# instead of an integer - and an incorrect end  - the end of the first
	# sub feature added, not of the union of all of them.

	# Also, the SeqIO::genbank.pm output is odd - the sub features appear
	# to be listed with the *previous* feature, not this one.

	for my $location (@locations) {
	    my $subgsf = $self->_parse_bsml_location($location);
	  #  print "start ", $subgsf->start,"\n";
	  #  print "end ", $subgsf->end,"\n";
	    $basegsf->add_sub_SeqFeature($subgsf, 'EXPAND');
	}
	# print $feat->getAttribute('id'),"\n";
	# print $basegsf->primary_tag,"\n";

    } else {
	# What to do if there are no locations? Nothing needed?
    }

    # Look at any <Attribute>s or <Qualifier>s that are present:
    my $floppies = &GETFLOPPIES($feat);
    for my $attr (@{$floppies}) {
	my ($name, $content) = &FLOPPYVALS($attr);
	# Don't know what the object is, dump it to a tag:
	$basegsf->add_tag_value(lc($name), $content);	
    }

    # Mostly this helps with debugging, but may be of utility...
    # Add a tag holding the BSML id value
    if ( (my $val = $feat->getAttribute('id')) &&
	 !$basegsf->has_tag('bsml-id')) {
	# Decided that this got a little sloppy...
#	$basegsf->add_tag_value("bsml-id", $val);
    }
    return $basegsf;
}

=head2 _parse_bsml_location

 Title   : _parse_bsml_location
 Usage   : $obj->_parse_bsml_feature( $intOrSiteLoc, $gsfObject )
 Function: Will examine the <Interval-loc> or <Site-loc> element provided
 Returns : Bio::SeqFeature::Generic
 Args    : 0 XML::DOM::Element <Interval/Site-loc> being analyzed.
           1 Optional SeqFeature::Generic to use

=cut

sub _parse_bsml_location {
    my $self = shift;
    my ($loc, $gsf) = @_;

    $gsf ||= Bio::SeqFeature::Generic->new();
    my $type = $loc->getNodeName;
    my ($start, $end);
    if ($type eq 'Interval-loc') {
	$start = $loc->getAttribute('startpos');
	$end = $loc->getAttribute('endpos');
    } elsif ($type eq 'Site-loc') {
	$start = $end = $loc->getAttribute('sitepos');
    } else {
	warn "Unknown location type '$type', could not make GSF\n";
	return;
    }
    $gsf->start($start);
    $gsf->end($end);

    # BSML does not have an explicit method to set undefined strand
    if (my $s = $loc->getAttribute("complement")) {
	if ($s) {
	    $gsf->strand(-1);
	} else {
	    $gsf->strand(1);
	}
    } else {
	# We're setting "strand nonspecific" here - bad idea?
	# In most cases the user likely meant it to be on the + strand
	$gsf->strand(0);
    }

    return $gsf;
}

=head2 _parse_reference

 Title   : _parse_reference
 Usage   : $obj->_parse_reference(@args )
 Function: Makes a new <Reference> object from a ::Reference, which is
           then stored in an array provide by -refs. It will be
           appended to the XML tree later.
 Returns :
 Args    : Argument array. Recognized keys:

      -xml The DOM::Document being modified

   -refobj The Annotation::Reference Object

     -refs An array reference to hold the new <Reference> DOM object

       -id Optional. If the XML id for the 'calling' element is
           provided, it will be placed in any <Reference> refs
           attribute.

=cut

sub _parse_reference {
    my $self = shift;
    my $args = $self->_parseparams( @_);
    my ($xml, $ref, $refRef) = ($args->{XML}, $args->{REFOBJ}, $args->{REFS});

    ###############
    # <Reference> #
    ###############

    my $xmlRef = $xml->createElement("Reference");
#>> This may not be the right way to make a BSML dbxref...
    if (my $link = $ref->medline) {
	$xmlRef->setAttribute('dbxref', $link);
    }

    # Make attributes for some of the characteristics
    my %stuff = ( start => $ref->start,
		  end => $ref->end,
		  rp => $ref->rp,
		  comment => $ref->comment,
		  pubmed => $ref->pubmed,
		  );
    for my $s (keys %stuff) {
	$self->_addel($xmlRef, "Attribute", {
	    name => $s, content => $stuff{$s} }) if ($stuff{$s});
    }
    $xmlRef->setAttribute('refs', $args->{ID}) if ($args->{ID});
    # Add the basic information
    # Should probably check for content before creation...
    $self->_addel($xmlRef, "RefAuthors")->
	appendChild( $xml->createTextNode(&STRIP($ref->authors)) );
    $self->_addel($xmlRef, "RefTitle")->
	appendChild( $xml->createTextNode(&STRIP($ref->title)) );
    $self->_addel($xmlRef, "RefJournal")->
	appendChild( $xml->createTextNode(&STRIP($ref->location)) );
    # References will be added later in a <Feature-Table>
    push @{$refRef}, $xmlRef;
}

=head2 _parse_annotation

 Title   : _parse_annotation
 Usage   : $obj->_parse_annotation(@args )
 Function: Will examine any Annotations found in -obj. Data found in
           ::Comment and ::DBLink structures, as well as Annotation
           description fields are stored in -desc for later
           generation of <Attribute>s. <Reference> objects are generated
           from ::References, and are stored in -refs - these will
           be appended to the XML tree later.
 Returns :
 Args    : Argument array. Recognized keys:

      -xml The DOM::Document being modified

      -obj Reference to the Bio object being analyzed

    -descr An array reference for holding description text items

     -refs An array reference to hold <Reference> DOM objects

       -id Optional. If the XML id for the 'calling' element is
           provided, it will be placed in any <Reference> refs
           attribute.

=cut

sub _parse_annotation {
    my $self = shift;
    my $args = $self->_parseparams( @_);
    my ($xml, $obj, $descRef, $refRef) =
	( $args->{XML}, $args->{OBJ}, $args->{DESC}, $args->{REFS} );
    # No good place to put any of this (except for references). Most stuff
    # just gets dumped to <Attribute>s
    my $ann = $obj->annotation;
    return  unless ($ann);
#	use BMS::Branch; my $debug = BMS::Branch->new( ); warn "$obj :"; $debug->branch($ann);
    unless (ref($ann) =~ /Collection/) {
	# Old style annotation. It seems that Features still use this
	# form of object
	$self->_parse_annotation_old(@_);
	return;
    }

    for my $key ($ann->get_all_annotation_keys()) {
	for my $thing ($ann->get_Annotations($key)) {
	    if ($key eq 'description') {
		push @{$descRef}, ["description" , $thing->value];
	    } elsif ($key eq 'comment') {
		push @{$descRef}, ["comment" , $thing->text];
	    } elsif ($key eq 'dblink') {
		# DBLinks get dumped to attributes, too
		push @{$descRef}, ["db_xref" ,  $thing->database . ":"
				   . $thing->primary_id ];
		if (my $com = $thing->comment) {
		    push @{$descRef}, ["link" , $com->text ];
		}

	    } elsif ($key eq 'reference') {
		$self->_parse_reference( @_, -refobj => $thing );
	    } elsif (ref($thing) =~ /SimpleValue/) {
		push @{$descRef}, [$key , $thing->value];
	    } else {
		# What is this??
		push @{$descRef}, ["error", "bsml.pm did not understand ".
				   "'$key' = '$thing'" ];
	    }
	}
    }
}

=head2 _parse_annotation_old

    Title   : _parse_annotation_old
 Usage   : $obj->_parse_annotation_old(@args)
 Function: As above, but for the old Annotation system.
           Apparently needed because Features are still using the old-style
           annotations?
 Returns :
 Args    : Argument array. Recognized keys:

      -xml The DOM::Document being modified

      -obj Reference to the Bio object being analyzed

    -descr An array reference for holding description text items

     -refs An array reference to hold <Reference> DOM objects

       -id Optional. If the XML id for the 'calling' element is
           provided, it will be placed in any <Reference> refs
           attribute.

=cut

    ###############
    # <Reference> #
    ###############

sub _parse_annotation_old {
    my $self = shift;
    my $args = $self->_parseparams( @_);
    my ($xml, $obj, $descRef, $refRef) =
	( $args->{XML}, $args->{OBJ}, $args->{DESC}, $args->{REFS} );
    # No good place to put any of this (except for references). Most stuff
    # just gets dumped to <Attribute>s
    if (my $ann = $obj->annotation) {
	push @{$descRef}, ["annotation", $ann->description];
	for my $com ($ann->each_Comment) {
	    push @{$descRef}, ["comment" , $com->text];
	}

	# Gene names just get dumped to <Attribute name="gene">
	for my $gene ($ann->each_gene_name) {
	    push @{$descRef}, ["gene" , $gene];
	}

	# DBLinks get dumped to attributes, too
	for my $link ($ann->each_DBLink) {
	    push @{$descRef}, ["db_xref" ,
			       $link->database . ":" . $link->primary_id ];
	    if (my $com = $link->comment) {
		push @{$descRef}, ["link" , $com->text ];
	    }
	}

	# References get produced and temporarily held
	for my $ref ($ann->each_Reference) {
	    $self->_parse_reference( @_, -refobj => $ref );
	}
    }
}

=head2 _add_page

 Title   : _add_page
 Usage   : $obj->_add_page($xmlDocument, $xmlSequenceObject)
 Function: Adds a simple <Page> and <View> structure for a <Sequence>
 Returns : a reference to the newly created <Page>
 Args    : 0 The DOM::Document being modified
           1 Reference to the <Sequence> object

=cut

sub _add_page {
    my $self = shift;
    my ($xml, $seq) = @_;
    my $disp = $xml->getElementsByTagName("Display")->item(0);
    my $page = $self->_addel($disp, "Page");
    my ($width, $height) = ( 7.8, 5.5);
    my $screen = $self->_addel($page, "Screen", {
	width => $width, height => $height, });
#    $screen->appendChild($xml->createComment("Must close explicitly"));
    my $view = $self->_addel($page, "View", {
	seqref => $seq->getAttribute('id'),
	title => $seq->getAttribute('title'),
	title1 => "{NAME}",
	title2 => "{LENGTH} {UNIT}",
    });
    $self->_addel($view, "View-line-widget", {
	shape => 'horizontal',
	hcenter => $width/2 + 0.7,
	'linear-length' => $width - 2,
    });
    $self->_addel($view, "View-axis-widget");
    return $page;
}


=head2 _addel

 Title   : _addel
 Usage   : $obj->_addel($parentElem, 'ChildName',
			{ anAttr => 'someValue', anotherAttr => 'aValue',})
 Function: Add an element with attribute values to a DOM tree
 Returns : a reference to the newly added element
 Args    : 0 The DOM::Element parent that you want to add to
           1 The name of the new child element
           2 Optional hash reference containing
             attribute name => attribute value assignments

=cut

sub _addel {
    my $self = shift;
    my ($root, $name, $attr) = @_;

    # Find the DOM::Document for the parent
    my $doc = $root->getOwnerDocument || $root;
    my $elem = $doc->createElement($name);
    for my $a (keys %{$attr}) {
	$elem->setAttribute($a, $attr->{$a});
    }
    $root->appendChild($elem);
    return $elem;
}

=head2 _show_dna

 Title   : _show_dna
 Usage   : $obj->_show_dna($newval)
 Function: (cut-and-pasted directly from embl.pm)
 Returns : value of _show_dna
 Args    : newvalue (optional)

=cut

sub _show_dna {
   my $obj = shift;
   if( @_ ) {
      my $value = shift;
      $obj->{'_show_dna'} = $value;
    }
    return $obj->{'_show_dna'};
}

=head2 _initialize

 Title   : _initialize
 Usage   : $dom = $obj->_initialize(@args)
 Function: Coppied from embl.pm, and augmented with initialization of the
           XML DOM tree
 Returns :
 Args    : -file => the XML file to be parsed

=cut

sub _initialize {
  my($self,@args) = @_;

  $self->SUPER::_initialize(@args);
  # hash for functions for decoding keys.
  $self->{'_func_ftunit_hash'} = {};
  $self->_show_dna(1); # sets this to one by default. People can change it

  my %param = @args;  # From SeqIO.pm
  @param{ map { lc $_ } keys %param } = values %param; # lowercase keys
  if ( exists $param{-file} && $param{-file} !~ /^>/) {
      # Is it blasphemy to add your own keys to an object in another package?
      # domtree => the parsed DOM tree retruned by XML::DOM
      $self->{'domtree'} = $self->_parse_xml( $param{-file} );
      # current_node => the <Sequence> node next in line for next_seq
      $self->{'current_node'} = 0;
  }

  $self->sequence_factory( Bio::Seq::SeqFactory->new
			   ( -verbose => $self->verbose(),
			     -type => 'Bio::Seq::RichSeq'))
      if( ! defined $self->sequence_factory );
}


=head2 _parseparams

 Title   : _parseparams
 Usage   : my $paramHash = $obj->_parseparams(@args)
 Function: Borrowed from Bio::Parse.pm, who borrowed it from CGI.pm
           Lincoln Stein -> Richard Resnick -> here
 Returns : A hash reference of the parameter keys (uppercase) pointing to
           their values.
 Args    : An array of key, value pairs. Easiest to pass values as:
           -key1 => value1, -key2 => value2, etc
           Leading "-" are removed.

=cut

sub _parseparams {
    my $self = shift;
    my %hash = ();
    my @param = @_;

    # Hacked out from Parse.pm
    # The next few lines strip out the '-' characters which
    # preceed the keys, and capitalizes them.
    for (my $i=0;$i<@param;$i+=2) {
        $param[$i]=~s/^\-//;
        $param[$i]=~tr/a-z/A-Z/;
    }
    pop @param if @param %2;  # not an even multiple
    %hash = @param;
    return \%hash;
}

=head2 _parse_xml

 Title   : _parse_xml
 Usage   : $dom = $obj->_parse_xml($filename)
 Function: uses XML::DOM to construct a DOM tree from the BSML document
 Returns : a reference to the parsed DOM tree
 Args    : 0 Path to the XML file needing to be parsed

=cut

sub _parse_xml {
    my $self = shift;
    my $file = shift;

    unless (-e $file) {
	$self->throw("Could not parse non-existant XML file '$file'.");
	return;
    }
    my $parser = XML::DOM::Parser->new();
    my $doc = $parser->parsefile ($file);
    return $doc;
}

sub DESTROY {
    my $self = shift;
    # Reports off the net imply that DOM::Parser will memory leak if you
    # do not explicitly dispose of it:
    # http://aspn.activestate.com/ASPN/Mail/Message/perl-xml/788458
    my $dom = $self->{'domtree'};
    # For some reason the domtree can get undef-ed somewhere...
    $dom->dispose if ($dom);
}


=head1 TESTING SCRIPT

 The following script may be used to test the conversion process. You
 will need a file of the format you wish to test. The script will
 convert the file to BSML, store it in /tmp/bsmltemp, read that file
 into a new SeqIO stream, and write it back as the original
 format. Comparison of this second file to the original input file
 will allow you to track where data may be lost or corrupted. Note
 that you will need to specify $readfile and $readformat.

 use Bio::SeqIO;
 # Tests preservation of details during round-trip conversion:
 # $readformat -> BSML -> $readformat
 my $tempspot = "/tmp/bsmltemp";  # temp folder to hold generated files
 my $readfile = "rps4y.embl";     # The name of the file you want to test
 my $readformat = "embl";         # The format of the file being tested

 system "mkdir $tempspot" unless (-d $tempspot);
 # Make Seq object from the $readfile
 my $biostream = Bio::SeqIO->new( -file => "$readfile" );
 my $seq = $biostream->next_seq();

 # Write BSML from SeqObject
 my $bsmlout = Bio::SeqIO->new( -format => 'bsml',
				   -file => ">$tempspot/out.bsml");
 warn "\nBSML written to $tempspot/out.bsml\n";
 $bsmlout->write_seq($seq);
 # Need to kill object for following code to work... Why is this so?
 $bsmlout = "";

 # Make Seq object from BSML
 my $bsmlin = Bio::SeqIO->new( -file => "$tempspot/out.bsml",
				  -format => 'bsml');
 my $seq2 = $bsmlin->next_seq();

 # Write format back from Seq Object
 my $genout = Bio::SeqIO->new( -format => $readformat,
				   -file => ">$tempspot/out.$readformat");
 $genout->write_seq($seq2);
 warn "$readformat  written to $tempspot/out.$readformat\n";

 # BEING LOST:
 # Join information (not possible in BSML 2.2)
 # Sequence type (??)

=cut


1;