transcript fetching optimisations, allows optional immediate exon loading as...

transcript fetching optimisations, allows optional immediate exon loading as opposed to lazy-loading

modules/Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm

@@ -60,6 +60,8 @@ use Bio::EnsEMBL::Translation;
 
 use Bio::EnsEMBL::Utils::Exception qw( deprecate throw warning );
 
+use Time::HiRes qw(time);
+
 @ISA = qw( Bio::EnsEMBL::DBSQL::BaseFeatureAdaptor );
 
 
@@ -272,6 +274,109 @@ sub fetch_all_by_Gene {
 
 
 
+=head2 fetch_all_by_Slice
+
+  Arg [1]    : Bio::EnsEMBL::Slice $slice
+               The slice to fetch transcripts on.
+  Arg [3]    : (optional) boolean $load_exons
+               if true, exons will be loaded immediately rather than
+               lazy loaded later.
+  Example    : $transcripts = $
+  Description: Overrides superclass method to optionally load exons
+               immediately rather than lazy-loading them later.  This
+               is more efficient when there are a lot of transcripts whose
+               exons are going to be used.
+  Returntype : reference to list of transcripts
+  Exceptions : thrown if exon cannot be placed on transcript slice
+  Caller     : Slice::get_all_Transcripts
+
+=cut
+
+sub fetch_all_by_Slice {
+  my $self  = shift;
+  my $slice = shift;
+  my $load_exons = shift;
+
+  my $transcripts = $self->SUPER::fetch_all_by_Slice($slice);
+
+  # if there are 0 or 1 transcripts still do lazy-loading
+  if(!$load_exons || @$transcripts < 2) {
+    return $transcripts;
+  }
+
+  # preload all of the exons now, instead of lazy loading later
+  # faster than 1 query per transcript
+
+  # get extent of region spanned by transcripts
+  my ($min_start, $max_end);
+  foreach my $tr (@$transcripts) {
+    if(!defined($min_start) || $tr->start() < $min_start) {
+      $min_start = $tr->start();
+    }
+    if(!defined($max_end) || $tr->end() > $max_end) {
+      $max_end   = $tr->end();
+    }
+  }
+
+  $min_start += $slice->start() - 1;
+  $max_end   += $slice->start() - 1;
+
+  my $ext_slice;
+
+  if($min_start >= $slice->start() && $max_end <= $slice->end()) {
+    $ext_slice = $slice;
+  } else {
+    my $sa = $self->db()->get_SliceAdaptor();
+    my $ext_slice = $sa->fetch_by_region
+      ($slice->coord_system->name(), $slice->seq_region_name(),
+       $min_start,$max_end, $slice->strand(), $slice->coord_system->version());
+  }
+
+  # associate exon identifiers with transcripts
+
+  my %tr_hash = map {$_->dbID => $_} @$transcripts;
+
+  my $tr_id_str = '(' . join(',', keys %tr_hash) . ')';
+
+  my $sth = $self->prepare("SELECT transcript_id, exon_id, rank " .
+                           "FROM   exon_transcript " .
+                           "WHERE  transcript_id IN $tr_id_str");
+
+  $sth->execute();
+
+  my ($ex_id, $tr_id, $rank);
+  $sth->bind_columns(\$tr_id, \$ex_id, \$rank);
+
+  my %ex_tr_hash;
+
+  while($sth->fetch()) {
+    $ex_tr_hash{$ex_id} ||= [];
+    push @{$ex_tr_hash{$ex_id}}, [$tr_hash{$tr_id}, $rank];
+  }
+
+  $sth->finish();
+
+  my $ea = $self->db()->get_ExonAdaptor();
+  my $exons = $ea->fetch_all_by_Slice($ext_slice);
+
+  # move exons onto transcript slice, and add them to transcripts
+  foreach my $ex (@$exons) {
+    $ex = $ex->transfer($slice) if($slice != $ext_slice);
+
+    if(!$ex) {
+      throw("Unexpected. Exon could not be transfered onto transcript slice.");
+    }
+
+    foreach my $row (@{$ex_tr_hash{$ex->dbID()}}) {
+      my ($tr, $rank) = @$row;
+      $tr->add_Exon($ex, $rank);
+    }
+  }
+
+  return $transcripts;
+}
+
+
 
 =head2 fetch_all_by_external_name
 

modules/Bio/EnsEMBL/Slice.pm

@@ -1316,8 +1316,11 @@ sub get_all_Genes_by_type{
 
 =head2 get_all_Transcripts
 
-  Arg [1]    : none
-  Example    : @transcripts = @{$slice->get_all_Transcripts};
+  Arg [1]    : (optional) boolean $load_exons
+               If set to true exons will not be lazy-loaded but will instead
+               be loaded right away.  This is faster if the exons are
+               actually going to be used right away.
+  Example    : @transcripts = @{$slice->get_all_Transcripts)_};
   Description: Gets all transcripts which overlap this slice.  If you want to
                specify a particular analysis or type, then you are better off
                using get_all_Genes or get_all_Genes_by_type and iterating
@@ -1330,13 +1333,15 @@ sub get_all_Genes_by_type{
 
 sub get_all_Transcripts {
   my $self = shift;
+  my $load_exons = shift;
 
   if(!$self->adaptor()) {
     warning('Cannot get Transcripts without attached adaptor');
     return [];
   }
 
-  return $self->adaptor->db->get_TranscriptAdaptor->fetch_all_by_Slice($self);
+  my $ta = $self->adaptor()->db()->get_TranscriptAdaptor();
+  return $ta->fetch_all_by_Slice($self, $load_exons);
 }
 
 

modules/Bio/EnsEMBL/Transcript.pm

@@ -791,7 +791,7 @@ sub add_Attributes {
 =cut
 
 sub add_Exon{
-  my ($self,$exon) = @_;
+  my ($self,$exon, $rank) = @_;
 
   #yup - we are going to be picky here...
   unless(defined $exon && ref $exon && $exon->isa("Bio::EnsEMBL::Exon") ) {
@@ -800,6 +800,11 @@ sub add_Exon{
 
   $self->{'_trans_exon_array'} ||= [];
 
+  if(defined($rank)) {
+    $self->{'_trans_exon_array'}->[$rank-1] = $exon;
+    return;
+  }
+
   my $was_added = 0;
 
   my $ea = $self->{'_trans_exon_array'};