biofetch_oracle documentation.

sub new {

  my $class = shift;
  my $args = shift;
  my $self  = $class->SUPER::new($args);
  my ($preferred) = rearrange(['PREFERRED_TAGS'],$args);
  $self->_preferred_tags($preferred?$preferred:\%default_preferred_tags);  # if the caller sent their own preferences, then use these, otherwise use defaults.

  my ($proxy) = rearrange(['PROXY'],$args);
  if ($proxy) {
    my @args = ref($proxy) ? @$proxy : eval $proxy;
    $self->{_proxy} =\@ args if @args;
  }
  $self;

}

sub segment {

  my $self = shift;
  my @segments = $self->SUPER::segment(@_);

  if (!@segments) {
    my $refclass = $self->refclass;

    my %args = $self->setup_segment_args(@_);
    if ($args{-class} && $args{-class} =~ /$refclass/oi) {
      return unless $self->load_from_embl('embl'=>$args{-name});
      @segments = $self->SUPER::segment(@_);
    } elsif ($args{-class} && $args{-class} =~ /refseq|swall|embl/i) { #hack to get refseq names
      return unless $self->load_from_embl(lc($args{-class})=>$args{-name});
      $args{-class} = $self->refclass;
      @segments = $self->SUPER::segment(%args);
    }
  }

  $self->_multiple_return_args(@segments);
}

# default is to return 'Sequence' as the class of all references

}

sub load_from_embl {

  my $self = shift;
  my $db   = shift;
  my $acc  = shift or $self->throw('Must provide an accession ID');

  my $biofetch;
  if ($self->{_biofetch}{$db}) {
    $biofetch = $self->{_biofetch}{$db};
  } else {
    $biofetch = $self->{_biofetch}{$db} = Bio::DB::BioFetch->new(-db=>$db);
    $biofetch->retrieval_type('tempfile');
    $biofetch->proxy(@{$self->{_proxy}}) if $self->{_proxy};
  }

  my $seq  = eval {$biofetch->get_Seq_by_id($acc)} or return;
  $self->_load_embl($acc,$seq);
  1;

}

sub _load_embl {

  my $self = shift;
  my $acc  = shift;
  my $seq  = shift;
  my $refclass = $self->refclass;
  my $locus    = $seq->id;

  # begin loading
  $self->setup_load();

  # first synthesize the entry for the top-level feature
  my @aliases;
  foreach ($seq->accession,$seq->get_secondary_accessions) {
    next if lc($_) eq lc($acc);
    push @aliases,[Alias => $_];
  }
  $self->load_gff_line(
		       {
			ref    => $acc,
			class  => $refclass,
			source => 'EMBL',
			method => 'origin',
			start  => 1,
			stop   => $seq->length,
			score  => undef,
			strand => '.',
			phase  => '.',
			gclass => $self->refclass,
			gname  => $acc,
			tstart => undef,
			tstop  => undef,
			attributes  => [[Note => $seq->desc],@aliases],
		       }
		      );
  # now load each feature in turn
  for my $feat ($seq->all_SeqFeatures) {
    my $attributes = $self->get_attributes($feat);
    my $name       = $self->guess_name($attributes);

    my $location = $feat->location;
    my @segments = map {[$_->start,$_->end,$_->seq_id]}
      $location->can('sub_Location') ? $location->sub_Location : $location;

    my $type     =  $feat->primary_tag eq 'CDS' ? 'mRNA'  : $feat->primary_tag;
    my $parttype =  $feat->primary_tag eq 'gene' ? 'exon' : $feat->primary_tag;

    if ($feat->primary_tag =~ /^(gene|CDS)$/) {
      $self->load_gff_line( {
			     ref    => $acc,
			     class  => $refclass,
			     source => 'EMBL',
			     method => $type,
			     start  => $location->start,
			     stop   => $location->end,
			     score  => $feat->score || undef,
			     strand => $feat->strand > 0 ? '+' : ($feat->strand < 0 ? '-' : '.'),
			     phase  => $feat->frame || '.',
			     gclass => $name->[0],
			     gname  => $name->[1],
			     tstart => undef,
			     tstop  => undef,
			     attributes  => $attributes,
			    }
			  );
      @$attributes = ();
    }

    for my $segment (@segments) {

      $self->load_gff_line( {
			     ref    => $segment->[2] eq $locus ? $acc : $segment->[2],
			     class  => $refclass,
			     source => 'EMBL',
			     method => $parttype,
			     start  => $segment->[0],
			     stop   => $segment->[1],
			     score  => $feat->score || undef,
			     strand => $feat->strand > 0 ? '+' : ($feat->strand < 0 ? '-' : '.'),
			     phase  => $feat->frame || '.',
			     gclass => $name->[0],
			     gname  => $name->[1],
			     tstart => undef,
			     tstop  => undef,
			     attributes  => $attributes,
			    }
			  );
    }

  }

  # finish loading
  $self->finish_load();

  # now load the DNA
  $self->load_sequence_string($acc,$seq->seq);

  1;

}

sub guess_name {

  my $self = shift;
  my $attributes = shift;
# remove this fix when Lincoln fixes it properly
  return ["Misc" => "Misc"] unless ($attributes);  # these are arbitrary, and possibly destructive defaults
  my @ordered_attributes = sort {($self->_preferred_tags->{$a->[0]} || 0) <=> ($self->_preferred_tags->{$b->[0]} || 0)} @$attributes;
  my $best = pop @ordered_attributes;
  @$attributes = @ordered_attributes;
  return $best;
}



1;

}

_preferred_tags	No description	Code
new	Description	Code
segment	No description	Code
refclass	No description	Code
load_from_embl	No description	Code
load_from_file	No description	Code
_load_embl	No description	Code
get_attributes	No description	Code
guess_name	No description	Code