From 75b70e78a7693a78e201394e17deb4eea081346f Mon Sep 17 00:00:00 2001
From: Kevin Howe <klh@ebi.ac.uk>
Date: Thu, 8 Mar 2007 17:17:53 +0000
Subject: [PATCH] On storage of supporting features, check that the feature
isn't already present, and reuse it if it is.
---
modules/Bio/EnsEMBL/DBSQL/ExonAdaptor.pm | 35 +----
.../EnsEMBL/DBSQL/SupportingFeatureAdaptor.pm | 141 +++++++++++++++++
.../Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm | 36 +----
.../TranscriptSupportingFeatureAdaptor.pm | 142 ++++++++++++++++++
4 files changed, 287 insertions(+), 67 deletions(-)
diff --git a/modules/Bio/EnsEMBL/DBSQL/ExonAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/ExonAdaptor.pm
index c258ce69c5..6673f3d56b 100644
--- a/modules/Bio/EnsEMBL/DBSQL/ExonAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/ExonAdaptor.pm
@@ -299,39 +299,8 @@ sub store {
}
# Now the supporting evidence
- # should be stored from featureAdaptor
- my $sql = "insert into supporting_feature (exon_id, feature_id, feature_type)
- values(?, ?, ?)";
-
- my $sf_sth = $self->prepare($sql);
-
- my $anaAdaptor = $self->db->get_AnalysisAdaptor();
- my $dna_adaptor = $self->db->get_DnaAlignFeatureAdaptor();
- my $pep_adaptor = $self->db->get_ProteinAlignFeatureAdaptor();
- my $type;
-
- foreach my $sf (@{$exon->get_all_supporting_features}) {
- if(!$sf->isa("Bio::EnsEMBL::BaseAlignFeature")){
- throw("$sf must be an align feature otherwise" .
- "it can't be stored");
- }
-
- if($sf->isa("Bio::EnsEMBL::DnaDnaAlignFeature")){
- $dna_adaptor->store($sf);
- $type = 'dna_align_feature';
- }elsif($sf->isa("Bio::EnsEMBL::DnaPepAlignFeature")){
- $pep_adaptor->store($sf);
- $type = 'protein_align_feature';
- } else {
- warning("Supporting feature of unknown type. Skipping : [$sf]\n");
- next;
- }
-
- $sf_sth->bind_param(1, $exonId, SQL_INTEGER);
- $sf_sth->bind_param(2, $sf->dbID, SQL_INTEGER);
- $sf_sth->bind_param(3, $type, SQL_VARCHAR);
- $sf_sth->execute();
- }
+ my $esf_adaptor = $db->get_SupportingFeatureAdaptor;
+ $esf_adaptor->store($exonId, $exon->get_all_supporting_features);
#
# Finally, update the dbID and adaptor of the exon (and any component exons)
diff --git a/modules/Bio/EnsEMBL/DBSQL/SupportingFeatureAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/SupportingFeatureAdaptor.pm
index 53668130b1..0072171b33 100644
--- a/modules/Bio/EnsEMBL/DBSQL/SupportingFeatureAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/SupportingFeatureAdaptor.pm
@@ -95,5 +95,146 @@ sub fetch_all_by_Exon {
return $out;
}
+=head2 store
+ Arg [2] : Int $transID
+ The dbID of an EnsEMBL transcript to associate with supporting
+ features
+ Arg [1] : Ref to array of Bio::EnsEMBL::BaseAlignFeature (the support)
+ Example : $sfa->store($exon_id, \@features);
+ Description: Stores a set of alignment features and associates an EnsEMBL transcript
+ with them
+ Returntype : none
+ Exceptions : thrown when invalid dbID is passed to this method
+ Caller : TranscriptAdaptor
+ Status : Stable
+
+=cut
+
+sub store {
+ my ( $self, $exon_dbID, $aln_objs ) = @_;
+
+ my $pep_check_sql =
+ "SELECT protein_align_feature_id " .
+ "FROM protein_align_feature " .
+ "WHERE seq_region_id = ? " .
+ "AND seq_region_start = ? " .
+ "AND seq_region_end = ? " .
+ "AND seq_region_strand = ? " .
+ "AND hit_name = ? " .
+ "AND hit_start = ? " .
+ "AND hit_end = ? " .
+ "AND analysis_id = ? " .
+ "AND cigar_line = ? ";
+
+ my $dna_check_sql =
+ "SELECT dna_align_feature_id " .
+ "FROM dna_align_feature " .
+ "WHERE seq_region_id = ? " .
+ "AND seq_region_start = ? " .
+ "AND seq_region_end = ? " .
+ "AND seq_region_strand = ? " .
+ "AND hit_name = ? " .
+ "AND hit_start = ? " .
+ "AND hit_end = ? " .
+ "AND analysis_id = ? " .
+ "AND cigar_line = ? " .
+ "AND hit_strand = ? ";
+
+ my $assoc_check_sql =
+ "SELECT * " .
+ "FROM supporting_feature " .
+ "WHERE exon_id = $exon_dbID " .
+ "AND feature_type = ? " .
+ "AND feature_id = ? ";
+
+ my $assoc_write_sql = "INSERT into supporting_feature " .
+ "(exon_id, feature_id, feature_type) " .
+ "values(?, ?, ?)";
+
+ my $pep_check_sth = $self->prepare($pep_check_sql);
+ my $dna_check_sth = $self->prepare($dna_check_sql);
+ my $assoc_check_sth = $self->prepare($assoc_check_sql);
+ my $sf_sth = $self->prepare($assoc_write_sql);
+
+ my $dna_adaptor = $self->db->get_DnaAlignFeatureAdaptor();
+ my $pep_adaptor = $self->db->get_ProteinAlignFeatureAdaptor();
+
+ foreach my $f (@$aln_objs) {
+ # check that the feature is in toplevel coords
+
+ if($f->slice->start != 1 || $f->slice->strand != 1) {
+ #move feature onto a slice of the entire seq_region
+ my $tls = $self->db->get_sliceAdaptor->fetch_by_region($f->slice->coord_system->name(),
+ $f->slice->seq_region_name(),
+ undef, #start
+ undef, #end
+ undef, #strand
+ $f->slice->coord_system->version());
+ $f = $f->transfer($tls);
+
+ if(!$f) {
+ throw('Could not transfer Feature to slice of ' .
+ 'entire seq_region prior to storing');
+ }
+ }
+
+ if(!$f->isa("Bio::EnsEMBL::BaseAlignFeature")){
+ throw("$f must be an align feature otherwise" .
+ "it can't be stored");
+ }
+
+ my ($sf_dbID, $type, $adap, $check_sth);
+
+ my @check_args = ($self->db->get_SliceAdaptor->get_seq_region_id($f->slice),
+ $f->start,
+ $f->end,
+ $f->strand,
+ $f->hseqname,
+ $f->hstart,
+ $f->hend,
+ $f->analysis->dbID,
+ $f->cigar_string);
+
+ if($f->isa("Bio::EnsEMBL::DnaDnaAlignFeature")){
+ $adap = $dna_adaptor;
+ $check_sth = $dna_check_sth;
+ $type = 'dna_align_feature';
+ push @check_args, $f->hstrand;
+ } elsif($f->isa("Bio::EnsEMBL::DnaPepAlignFeature")){
+ $adap = $pep_adaptor;
+ $check_sth = $pep_check_sth;
+ $type = 'protein_align_feature';
+ } else {
+ warning("Supporting feature of unknown type. Skipping : [$f]\n");
+ next;
+ }
+
+ $check_sth->execute(@check_args);
+ $sf_dbID = $check_sth->fetchrow_array;
+ if (not $sf_dbID) {
+ $adap->store($f);
+ $sf_dbID = $f->dbID;
+ }
+
+ # now check association
+ $assoc_check_sth->execute($type,
+ $sf_dbID);
+ if (not $assoc_check_sth->fetchrow_array) {
+ $sf_sth->bind_param(1, $exon_dbID, SQL_INTEGER);
+ $sf_sth->bind_param(2, $sf_dbID, SQL_INTEGER);
+ $sf_sth->bind_param(3, $type, SQL_VARCHAR);
+ $sf_sth->execute();
+ }
+ }
+
+ $dna_check_sth->finish;
+ $pep_check_sth->finish;
+ $assoc_check_sth->finish;
+ $sf_sth->finish;
+
+}
+
+
+
1;
diff --git a/modules/Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm
index 4a8dc03732..c03d2ee767 100644
--- a/modules/Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/TranscriptAdaptor.pm
@@ -766,40 +766,8 @@ sub store {
}
# Now the supporting evidence
- # should be stored from featureAdaptor
- my $sql = "insert into transcript_supporting_feature
- (transcript_id, feature_id, feature_type)
- values(?, ?, ?)";
-
- my $sf_sth = $self->prepare($sql);
-
- my $anaAdaptor = $self->db->get_AnalysisAdaptor();
- my $dna_adaptor = $self->db->get_DnaAlignFeatureAdaptor();
- my $pep_adaptor = $self->db->get_ProteinAlignFeatureAdaptor();
- my $type;
-
- foreach my $sf (@{$transcript->get_all_supporting_features}) {
- if(!$sf->isa("Bio::EnsEMBL::BaseAlignFeature")){
- throw("$sf must be an align feature otherwise" .
- "it can't be stored");
- }
-
- if($sf->isa("Bio::EnsEMBL::DnaDnaAlignFeature")){
- $dna_adaptor->store($sf);
- $type = 'dna_align_feature';
- }elsif($sf->isa("Bio::EnsEMBL::DnaPepAlignFeature")){
- $pep_adaptor->store($sf);
- $type = 'protein_align_feature';
- } else {
- warning("Supporting feature of unknown type. Skipping : [$sf]\n");
- next;
- }
-
- $sf_sth->bind_param(1, $transc_dbID, SQL_INTEGER);
- $sf_sth->bind_param(2, $sf->dbID, SQL_INTEGER);
- $sf_sth->bind_param(3, $type, SQL_VARCHAR);
- $sf_sth->execute();
- }
+ my $tsf_adaptor = $db->get_TranscriptSupportingFeatureAdaptor;
+ $tsf_adaptor->store($transc_dbID, $transcript->get_all_supporting_features);
# store transcript attributes if there are any
my $attr_adaptor = $db->get_AttributeAdaptor();
diff --git a/modules/Bio/EnsEMBL/DBSQL/TranscriptSupportingFeatureAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/TranscriptSupportingFeatureAdaptor.pm
index 323b3d3f28..a50dc319bd 100644
--- a/modules/Bio/EnsEMBL/DBSQL/TranscriptSupportingFeatureAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/TranscriptSupportingFeatureAdaptor.pm
@@ -96,5 +96,147 @@ sub fetch_all_by_Transcript {
return $out;
}
+
+
+=head2 store
+ Arg [2] : Int $transID
+ The dbID of an EnsEMBL transcript to associate with supporting
+ features
+ Arg [1] : Ref to array of Bio::EnsEMBL::BaseAlignFeature (the support)
+ Example : $dbea->store($transcript_id, \@features);
+ Description: Stores a set of alignment features and associates an EnsEMBL transcript
+ with them
+ Returntype : none
+ Exceptions : thrown when invalid dbID is passed to this method
+ Caller : TranscriptAdaptor
+ Status : Stable
+
+=cut
+
+sub store {
+ my ( $self, $tran_dbID, $aln_objs ) = @_;
+
+ my $pep_check_sql =
+ "SELECT protein_align_feature_id " .
+ "FROM protein_align_feature " .
+ "WHERE seq_region_id = ? " .
+ "AND seq_region_start = ? " .
+ "AND seq_region_end = ? " .
+ "AND seq_region_strand = ? " .
+ "AND hit_name = ? " .
+ "AND hit_start = ? " .
+ "AND hit_end = ? " .
+ "AND analysis_id = ? " .
+ "AND cigar_line = ? ";
+
+ my $dna_check_sql =
+ "SELECT dna_align_feature_id " .
+ "FROM dna_align_feature " .
+ "WHERE seq_region_id = ? " .
+ "AND seq_region_start = ? " .
+ "AND seq_region_end = ? " .
+ "AND seq_region_strand = ? " .
+ "AND hit_name = ? " .
+ "AND hit_start = ? " .
+ "AND hit_end = ? " .
+ "AND analysis_id = ? " .
+ "AND cigar_line = ? " .
+ "AND hit_strand = ? ";
+
+ my $assoc_check_sql =
+ "SELECT * " .
+ "FROM transcript_supporting_feature " .
+ "WHERE transcript_id = $tran_dbID " .
+ "AND feature_type = ? " .
+ "AND feature_id = ? ";
+
+ my $assoc_write_sql = "INSERT into transcript_supporting_feature " .
+ "(transcript_id, feature_id, feature_type) " .
+ "values(?, ?, ?)";
+
+ my $pep_check_sth = $self->prepare($pep_check_sql);
+ my $dna_check_sth = $self->prepare($dna_check_sql);
+ my $assoc_check_sth = $self->prepare($assoc_check_sql);
+ my $sf_sth = $self->prepare($assoc_write_sql);
+
+ my $dna_adaptor = $self->db->get_DnaAlignFeatureAdaptor();
+ my $pep_adaptor = $self->db->get_ProteinAlignFeatureAdaptor();
+
+ foreach my $f (@$aln_objs) {
+ # check that the feature is in toplevel coords
+
+ if($f->slice->start != 1 || $f->slice->strand != 1) {
+ #move feature onto a slice of the entire seq_region
+ my $tls = $self->db->get_sliceAdaptor->fetch_by_region($f->slice->coord_system->name(),
+ $f->slice->seq_region_name(),
+ undef, #start
+ undef, #end
+ undef, #strand
+ $f->slice->coord_system->version());
+ $f = $f->transfer($tls);
+
+ if(!$f) {
+ throw('Could not transfer Feature to slice of ' .
+ 'entire seq_region prior to storing');
+ }
+ }
+
+ if(!$f->isa("Bio::EnsEMBL::BaseAlignFeature")){
+ throw("$f must be an align feature otherwise" .
+ "it can't be stored");
+ }
+
+ my ($sf_dbID, $type, $adap, $check_sth);
+
+ my @check_args = ($self->db->get_SliceAdaptor->get_seq_region_id($f->slice),
+ $f->start,
+ $f->end,
+ $f->strand,
+ $f->hseqname,
+ $f->hstart,
+ $f->hend,
+ $f->analysis->dbID,
+ $f->cigar_string);
+
+ if($f->isa("Bio::EnsEMBL::DnaDnaAlignFeature")){
+ $adap = $dna_adaptor;
+ $check_sth = $dna_check_sth;
+ $type = 'dna_align_feature';
+ push @check_args, $f->hstrand;
+ } elsif($f->isa("Bio::EnsEMBL::DnaPepAlignFeature")){
+ $adap = $pep_adaptor;
+ $check_sth = $pep_check_sth;
+ $type = 'protein_align_feature';
+ } else {
+ warning("Supporting feature of unknown type. Skipping : [$f]\n");
+ next;
+ }
+
+ $check_sth->execute(@check_args);
+ $sf_dbID = $check_sth->fetchrow_array;
+ if (not $sf_dbID) {
+ $adap->store($f);
+ $sf_dbID = $f->dbID;
+ }
+
+ # now check association
+ $assoc_check_sth->execute($type,
+ $sf_dbID);
+ if (not $assoc_check_sth->fetchrow_array) {
+ $sf_sth->bind_param(1, $tran_dbID, SQL_INTEGER);
+ $sf_sth->bind_param(2, $sf_dbID, SQL_INTEGER);
+ $sf_sth->bind_param(3, $type, SQL_VARCHAR);
+ $sf_sth->execute();
+ }
+ }
+
+ $dna_check_sth->finish;
+ $pep_check_sth->finish;
+ $assoc_check_sth->finish;
+ $sf_sth->finish;
+
+}
+
+
1;
--
GitLab