view release on metacpan or  search on metacpan

lib/Bio/AlignIO/stockholm.pm  view on Meta::CPAN

  database_comment
  custom
  dblink
  alignment_comment
  num_sequences
  seq_annotation
  );

# This maps the tagname back to a tagname-annotation value combination.
# Some data is stored using get/set methods ('Methods'), others
# are mapped b/c of more complex annotation types.

our %WRITEMAP = (
    'accession'             =>  'AC/Method',
    'id'                    =>  'ID/Method',
    'description'           =>  'DE/Method',
    'record_authors'        =>  'AU/SimpleValue',
    'seed_source'           =>  'SE/SimpleValue',
    'build_command'         =>  'BM/SimpleValue',
    'gathering_threshold'   =>  'GA/SimpleValue',
    'noise_cutoff'          =>  'NC/SimpleValue',
    'trusted_cutoff'        =>  'TC/SimpleValue',
    'entry_type'            =>  'TP/SimpleValue',
    'num_sequences'         =>  'SQ/SimpleValue',
    'previous_ids'          =>  'PI/SimpleValue',
    'database_comment'      =>  'DC/SimpleValue',
    'dblink'                =>  'DR/DBLink',
    'reference'             =>  'RX/Reference',
    'ref_number'            =>  'RN/number',
    'ref_comment'           =>  'RC/comment',
    'ref_pubmed'            =>  'RM/pubmed',
    'ref_title'             =>  'RT/title',
    'ref_authors'           =>  'RA/authors',
    'ref_location'          =>  'RL/location',
    'alignment_comment'     =>  'CC/Comment',
    'seq_annotation'        =>  'DR/Collection',
    #Pfam-specific
    'build_method'          =>  'AM/SimpleValue',
    'pfam_family_accession' =>  'NE/SimpleValue',
    'seq_start_stop'        =>  'NL/SimpleValue',
    # Rfam-specific GF lines
    'sec_structure_source'  =>  'SS/SimpleValue',
    # custom; this is used to carry over anything from the input alignment
    # not mapped to LocatableSeqs or SimpleAlign in a meaningful way
    'custom'                =>  'XX/SimpleValue'
);

# This maps the tagname back to a tagname-annotation value combination.
# Some data is stored using get/set methods ('Methods'), others
# are mapped b/c of more complex annotation types.

=head2 new

 Title   : new
 Usage   : my $alignio = Bio::AlignIO->new(-format => 'stockholm'
					  -file   => '>file');
 Function: Initialize a new L<Bio::AlignIO::stockholm> reader or writer
 Returns : L<Bio::AlignIO> object
 Args    : -line_length :  length of the line for the alignment block
           -alphabet    :  symbol alphabet to set the sequences to.  If not set,
                           the parser will try to guess based on the alignment
                           accession (if present), defaulting to 'dna'.
           -spaces      :  (optional, def = 1) boolean to add a space in between
                           the "# STOCKHOLM 1.0" header and the annotation and
                           the annotation and the alignment.

=cut

sub _initialize {
    my ( $self, @args ) = @_;
    $self->SUPER::_initialize(@args);
    my ($handler, $linelength, $spaces) = $self->_rearrange([qw(HANDLER LINE_LENGTH SPACES)],@args);
    $spaces = defined $spaces ? $spaces : 1;
    $self->spaces($spaces);
    # hash for functions for decoding keys.
    $handler ? $self->alignhandler($handler) :
    $self->alignhandler(Bio::AlignIO::Handler::GenericAlignHandler->new(
                    -format => 'stockholm',
                    -verbose => $self->verbose,
                    ));
    $linelength && $self->line_length($linelength);
}

=head2 next_aln

 Title   : next_aln
 Usage   : $aln = $stream->next_aln()
 Function: returns the next alignment in the stream.
 Returns : L<Bio::Align::AlignI> object
 Args    : NONE

=cut

sub next_aln {
    my $self = shift;

    my $handler = $self->alignhandler;
    # advance to alignment header
    while( defined(my $line = $self->_readline) ) {
        if ($line =~ m{^\#\s*STOCKHOLM\s+}xmso) {
            last;
        }
    }

    $self->{block_line} = 0;
    # go into main body of alignment
    my ($data_chunk, $isa_primary, $name, $alphabet);
    my $last_feat = '';
    while( defined(my $line = $self->_readline) ) {
        # only blank lines are in between blocks, so reset block line
        my ($primary_tag, $secondary_tag, $data, $nse, $feat, $align, $concat);
        if ($line =~ m{^\s*$}xmso) {
            $self->{block_line} &&= 0;
            next;
        }

        # End of Record
        if (index($line, '//') == 0) {
            # fencepost
            $handler->data_handler($data_chunk);
            undef $data_chunk;