kegg documentation.

This class transforms KEGG gene records into Bio::Seq objects. This section is supposed to document which sections and properties of
a KEGG databank record end up where in the Bioperl object model. It
is far from complete and presently focuses only on those mappings
which may be non-obvious. $seq in the text refers to the
Bio::Seq::RichSeqI implementing object returned by the parser for each
record.

'ENTRY'

$seq->primary_id

'NAME'

$seq->display_id

'DEFINITION'

$seq->annotation->get_Annotations('description');

'ORTHOLOG'

grep {$_->database eq 'KO'} $seq->annotation->get_Annotations('dblink')

'CLASS'

grep {$_->database eq 'PATH'}
$seq->annotation->get_Annotations('dblink')

'POSITION'

FIXME, NOT IMPLEMENTED

'PATHWAY'

for my $pathway ( $seq->annotation->get_Annotations('pathway') ) {
#
}

'DBLINKS'

$seq->annotation->get_Annotations('dblink')

'CODON_USAGE'

FIXME, NOT IMPLEMENTED

'AASEQ'

$seq->translate->seq

'NTSEQ'

$seq-E<gt>seq

sub next_seq {

	my ($self,@args) = @_;
	my $builder = $self->sequence_builder();
	my $seq;
	my %params;

	my $buffer;
	my (@acc, @features);
	my ($display_id, $annotation);
	my $species;

	# initialize; we may come here because of starting over
	@features = ();
	$annotation = undef;
	@acc = ();
	$species = undef;
	%params = (-verbose => $self->verbose); # reset hash
	local($/) = "///\n";

	$buffer = $self->_readline();

	return if( !defined $buffer ); # end of file
	$buffer =~ /^ENTRY/ ||
	  $self->throw("KEGG stream with bad ENTRY line. Not KEGG in my book. Got $buffer'");

	my %FIELDS;
	my @chunks = split /\n(?=\S)/, $buffer;

	foreach my $chunk (@chunks){
		my($key) = $chunk =~ /^(\S+)/;
		$FIELDS{$key} = $chunk;
	}

	# changing to split method to get entry_ids that include
	# sequence version like Whatever.1
	my(undef,$entry_id,$entry_seqtype,$entry_species) =
	  split(' ',$FIELDS{ENTRY});

	my($name);
	if ($FIELDS{NAME}) {
          ($name) = $FIELDS{NAME} =~ /^NAME\s+(.+)$/;
	}

        my( $definition, $aa_length, $aa_seq, $nt_length, $nt_seq );

        if(( exists $FIELDS{DEFINITION} ) and ( $FIELDS{DEFINITION} =~ /^DEFINITION/ )) {
          ($definition) = $FIELDS{DEFINITION} =~ /^DEFINITION\s+(.+)$/s;
          $definition =~ s/\s+/ /gs;
        }
        if(( exists $FIELDS{AASEQ} ) and ( $FIELDS{AASEQ} =~ /^AASEQ/ )) {
          ($aa_length,$aa_seq) = $FIELDS{AASEQ} =~ /^AASEQ\s+(\d+)\n(.+)$/s;
          $aa_seq =~ s/\s+//g;
        }
        if(( exists  $FIELDS{NTSEQ} ) and ( $FIELDS{NTSEQ} =~ /^NTSEQ/ )) {
          ($nt_length,$nt_seq) = $FIELDS{NTSEQ} =~ /^NTSEQ\s+(\d+)\n(.+)$/s;
          $nt_seq =~ s/\s+//g;
        }

	$annotation = Bio::Annotation::Collection->new();

	$annotation->add_Annotation('description',
						Bio::Annotation::Comment->new(-text => $definition));

	$annotation->add_Annotation('aa_seq',
						Bio::Annotation::Comment->new(-text => $aa_seq));

	my($ortholog_db,$ortholog_id,$ortholog_desc);
	if ($FIELDS{ORTHOLOG}) {
		($ortholog_db,$ortholog_id,$ortholog_desc) = $FIELDS{ORTHOLOG}
		  =~ /^ORTHOLOG\s+(\S+):\s+(\S+)\s+(.*?)$/;

        $annotation->add_Annotation('dblink',Bio::Annotation::DBLink->new(
                     -database => $ortholog_db,
                     -primary_id => $ortholog_id,
                     -comment => $ortholog_desc) );
  }

  if($FIELDS{MOTIF}){
     $FIELDS{MOTIF} =~ s/^MOTIF\s+//;
     while($FIELDS{MOTIF} =~/\s*?(\S+):\s+(.+?)$/mg){
         my $db = $1;
         my $ids = $2;
         foreach my $id (split(/\s+/, $ids)){

     $annotation->add_Annotation('dblink',Bio::Annotation::DBLink->new(
              -database =>$db,
              -primary_id => $id,
              -comment => "")   );
        }
     }
  }

  if($FIELDS{PATHWAY}) {
     $FIELDS{PATHWAY} =~ s/^PATHWAY\s+//;
     while($FIELDS{PATHWAY} =~ /\s*PATH:\s+(.+)$/mg){
        $annotation->add_Annotation('pathway',
           Bio::Annotation::Comment->new(-text => "$1"));
     }
  }

  if ($FIELDS{CLASS}) {
      $FIELDS{CLASS} =~ s/^CLASS\s+//;
      $FIELDS{'CLASS'} =~ s/\n//g;
      while($FIELDS{CLASS} =~ /(.*?)\[(\S+):(\S+)\]/g){
          my ($pathway,$db,$id) = ($1,$2,$3);
          $pathway =~ s/\s+/ /g;
          $pathway =~ s/\s$//g;
          $pathway =~ s/^\s+//;
          $annotation->add_Annotation('pathway',
                  Bio::Annotation::Comment->new(-text => $pathway));

          $annotation->add_Annotation('dblink',Bio::Annotation::DBLink->new(
                      -database => $db, -primary_id => $id));
      }
  }

  if($FIELDS{DBLINKS}) {
      $FIELDS{DBLINKS} =~ s/^DBLINKS/       /;
      while($FIELDS{DBLINKS} =~ /\s+(\S+):\s+(\S+)\n?/gs){ ### modified
           $annotation->add_Annotation('dblink',Bio::Annotation::DBLink->new(
                    -database => $1, -primary_id => $2)) if $1;
      }
  }

  $params{'-alphabet'}         = 'dna';
  $params{'-seq'}              = $nt_seq;
  $params{'-display_id'}       = $name;
  $params{'-accession_number'} = $entry_id;
  $params{'-species'}          = Bio::Species->new(
											  -common_name => $entry_species);
  $params{'-annotation'}       = $annotation;

  $builder->add_slot_value(%params);
  $seq = $builder->make_object();

  return $seq;

}

_initialize	No description	Code
next_seq	Description	Code