Merge 3d64a02 into 5308e4c

.gitignore

@@ -32,6 +32,7 @@ a.out
 *.obj
 *.class
 MultiTestDB.conf
+MultiTestDB.*.conf
 *.frozen.conf
 bioperl*.zip
 /modules/t/CLEAN.t

modules/Bio/EnsEMBL/Biotype.pm

@@ -86,6 +86,8 @@ use parent qw(Bio::EnsEMBL::Storable);
       string - the name of the biotype group (for ensembl)
   Arg [-SO_ACC] :
       string - the Sequence Ontology accession of this biotype
+  Arg [-SO_TERM] :
+      string - the Sequence Ontology term for the SO accession of this biotype
   Arg [-DESCRIPTION] :
       string - the biotype description
   Arg [-DB_TYPE] :
@@ -107,14 +109,15 @@ sub new {
 
   my $self = $class->SUPER::new();
 
-  my($dbID, $name, $object_type, $biotype_group, $so_acc, $description, $db_type, $attrib_type_id) =
-    rearrange([qw(BIOTYPE_ID NAME OBJECT_TYPE BIOTYPE_GROUP SO_ACC DESCRIPTION DB_TYPE ATTRIB_TYPE_ID)], @args);
+  my($dbID, $name, $object_type, $biotype_group, $so_acc, $so_term, $description, $db_type, $attrib_type_id) =
+    rearrange([qw(BIOTYPE_ID NAME OBJECT_TYPE BIOTYPE_GROUP SO_ACC SO_TERM DESCRIPTION DB_TYPE ATTRIB_TYPE_ID)], @args);
 
   $self->{'dbID'} = $dbID;
   $self->{'name'} = $name;
   $self->{'object_type'} = $object_type;
   $self->{'biotype_group'} = $biotype_group;
   $self->{'so_acc'} = $so_acc;
+  $self->{'so_term'} = $so_term;
   $self->{'description'} = $description;
   $self->{'db_type'} = $db_type;
   $self->{'attrib_type_id'} = $attrib_type_id;
@@ -215,6 +218,26 @@ sub so_acc {
   return $self->{'so_acc'};
 }
 
+=head2 so_term
+
+  Arg [1]    : (optional) string $so_term
+  Example    : $feat->so_term();
+  Description: Getter/Setter for the Sequence Ontology term of this biotype.
+  Returntype : string
+  Exceptions : none
+
+=cut
+
+sub so_term {
+  my ( $self, $so_term ) = @_;
+
+  if ( defined($so_term) ) {
+    $self->{'so_term'} = $so_term;
+  }
+
+  return $self->{'so_term'};
+}
+
 =head2 object_type
 
   Arg [1]    : (optional) string $object_type

modules/Bio/EnsEMBL/CDS.pm

@@ -66,7 +66,10 @@ use Scalar::Util qw(weaken isweak);
 
 @ISA = qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0000316';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000316',
+  term => 'CDS',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/DBSQL/BiotypeAdaptor.pm

@@ -90,7 +90,7 @@ sub _tables {
 sub _columns {
   my $self = shift;
 
-  return ('b.biotype_id', 'b.name', 'b.object_type', 'b.db_type', 'b.attrib_type_id', 'b.description', 'b.biotype_group', 'b.so_acc');
+  return ('b.biotype_id', 'b.name', 'b.object_type', 'b.db_type', 'b.attrib_type_id', 'b.description', 'b.biotype_group', 'b.so_acc', 'b.so_term');
 }
 
 =head2 _objs_from_sth
@@ -107,9 +107,9 @@ sub _columns {
 sub _objs_from_sth {
   my ($self, $sth) = @_;
 
-  my ($dbID, $name, $object_type, $db_type, $attrib_type_id, $description, $biotype_group, $so_acc);
+  my ($dbID, $name, $object_type, $db_type, $attrib_type_id, $description, $biotype_group, $so_acc, $so_term);
 
-  $sth->bind_columns(\$dbID, \$name, \$object_type, \$db_type, \$attrib_type_id, \$description, \$biotype_group, \$so_acc);
+  $sth->bind_columns(\$dbID, \$name, \$object_type, \$db_type, \$attrib_type_id, \$description, \$biotype_group, \$so_acc, \$so_term);
 
   my @biotypes;
 
@@ -124,6 +124,7 @@ sub _objs_from_sth {
          'description'    => $description,
          'biotype_group'  => $biotype_group,
          'so_acc'         => $so_acc,
+         'so_term'        => $so_term,
       } )
     );
   }

modules/Bio/EnsEMBL/DnaDnaAlignFeature.pm

@@ -54,7 +54,10 @@ use Bio::EnsEMBL::Utils::Exception qw(throw);
 
 @ISA = qw( Bio::EnsEMBL::BaseAlignFeature );
 
-use constant SO_ACC => 'SO:0000347';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000347',
+  term => 'nucleotide_match',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/DnaPepAlignFeature.pm

@@ -51,7 +51,10 @@ use vars qw(@ISA);
 
 @ISA = qw( Bio::EnsEMBL::BaseAlignFeature );
 
-use constant SO_ACC => 'SO:0000349';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000349',
+  term => 'protein_match',
+};
 
 =head2 _hit_unit
 

modules/Bio/EnsEMBL/Exon.pm

@@ -80,8 +80,10 @@ use Bio::EnsEMBL::DBSQL::SupportingFeatureAdaptor;
 use vars qw(@ISA);
 @ISA = qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0000147';
-
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000147',
+  term => 'exon',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/Feature.pm

@@ -1449,9 +1449,9 @@ sub get_nearest_Gene {
 =head2 feature_so_acc
 
   Description: This method returns a string containing the SO accession number of the feature
-               Define constant SO_ACC in classes that require it, or override it for multiple possible values for a class.
+               Define constant SEQUENCE_ONTOLOGY in classes that require it, or override it for multiple possible values for a class.
   Returntype : String (Sequence Ontology accession number)
-  Exceptions : Thrown if caller SO_ACC is undefined and is not a Bio::EnsEMBL::Feature
+  Exceptions : Thrown if caller SEQUENCE_ONTOLOGY is undefined and is not a Bio::EnsEMBL::Feature
 
 =cut
 
@@ -1463,16 +1463,43 @@ sub feature_so_acc {
 
   # Get the caller class SO acc
   try {
-    $so_acc = $ref->SO_ACC;
+    $so_acc = $ref->SEQUENCE_ONTOLOGY->{'acc'};
   };
- 
-  unless ($so_acc || $ref eq 'Bio::EnsEMBL::Feature' ) {
-    throw( "constant SO_ACC in ${ref} is not defined");
+
+  if (!$so_acc && $ref ne 'Bio::EnsEMBL::Feature' ) {
+    throw( "constant SEQUENCE_ONTOLOGY in ${ref} is not defined");
   }
 
   return $so_acc // 'SO:0000001';
 }
 
+=head2 feature_so_term
+
+  Description: This method returns a string containing the SO term of the feature
+               Define constant SEQUENCE_ONTOLOGY in classes that require it, or override it for multiple possible values for a class.
+  Returntype : String (Sequence Ontology term)
+  Exceptions : Thrown if caller SEQUENCE_ONTOLOGY is undefined and is not a Bio::EnsEMBL::Feature
+
+=cut
+
+sub feature_so_term {
+  my ($self) = @_;
+
+  my $ref = ref $self;
+  my $so_term;
+
+  # Get the caller class SO acc
+  try {
+    $so_term = $ref->SEQUENCE_ONTOLOGY->{'term'};
+  };
+
+  if (!$so_term && $ref ne 'Bio::EnsEMBL::Feature' ) {
+    throw( "constant SEQUENCE_ONTOLOGY in ${ref} is not defined");
+  }
+
+  return $so_term // 'region';
+}
+
 =head2 summary_as_hash
 
   Example       : $feature_summary = $feature->summary_as_hash();

modules/Bio/EnsEMBL/Gene.pm

@@ -73,7 +73,10 @@ use Bio::EnsEMBL::Utils::Scalar qw(assert_ref);
 
 use parent qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0000704';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000704',
+  term => 'gene',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/KaryotypeBand.pm

@@ -82,7 +82,10 @@ use Bio::EnsEMBL::Utils::Exception qw(warning);
 
 @ISA = qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0000341';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000341',
+  term => 'chromosome_band',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/MiscFeature.pm

@@ -125,7 +125,10 @@ use vars qw(@ISA);
 
 @ISA = qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0001411';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0001411',
+  term => 'biological_region',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/RepeatFeature.pm

@@ -82,7 +82,10 @@ use Bio::EnsEMBL::Utils::Argument qw(rearrange);
 
 use base qw/Bio::EnsEMBL::Feature/;
 
-use constant SO_ACC => 'SO:0000657';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000657',
+  term => 'repeat_region',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/SimpleFeature.pm

@@ -69,7 +69,10 @@ use Scalar::Util qw(weaken isweak);
 
 @ISA = qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0001411';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0001411',
+  term => 'biological_region',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/Transcript.pm

@@ -77,7 +77,10 @@ use Bio::EnsEMBL::Utils::Scalar qw( assert_ref );
 
 use parent qw(Bio::EnsEMBL::Feature);
 
-use constant SO_ACC => 'SO:0000673';
+use constant SEQUENCE_ONTOLOGY => {
+  acc  => 'SO:0000673',
+  term => 'transcript',
+};
 
 =head2 new
 

modules/Bio/EnsEMBL/UTR.pm

@@ -223,8 +223,14 @@ sub type {
 
 # package variable to minimize duplication
 my %utr_type_so_mapping = (
- 'five_prime_utr'  => 'SO:0000204',
- 'three_prime_utr' => 'SO:0000205'
+ 'five_prime_utr'  => {
+    acc  => 'SO:0000204',
+    term => 'five_prime_UTR',
+  },
+ 'three_prime_utr' => {
+    acc  => 'SO:0000205',
+    term => 'three_prime_UTR',
+  }
 );
 
 =head2 feature_so_acc
@@ -240,7 +246,23 @@ sub feature_so_acc {
   my $self = shift;
 
   # return UTR type SO acc, or UTR acc
-  return $utr_type_so_mapping{$self->type} // 'SO:0000203';
+  return $utr_type_so_mapping{$self->type}->{'acc'} // 'SO:0000203';
+}
+
+=head2 feature_so_term
+
+  Example    : print $utr->feature_so_term;
+  Description: This method returns a string containing the SO accession term of the UTR, based on type.
+               Overrides Bio::EnsEMBL::Feature::feature_so_term
+  Returntype : string (Sequence Ontology term)
+
+=cut
+
+sub feature_so_term {
+  my $self = shift;
+
+  # return UTR type SO acc, or UTR acc
+  return $utr_type_so_mapping{$self->type}->{'term'} // 'UTR';
 }
 
 =head2 summary_as_hash

modules/t/biotype.t

@@ -52,19 +52,21 @@ is($biotype1->object_type, 'gene', 'Biotype is from Gene object');
 is($biotype1->name, 'protein_coding', 'Biotype name is protein_coding');
 is($biotype1->biotype_group, 'coding', 'Biotype group is coding');
 is($biotype1->so_acc, 'SO:0001217', 'Biotype protein_coding refers to SO:0001217');
+is($biotype1->so_term, 'protein_coding_gene', 'Biotype protein_coding refers to SO term protein_coding_gene');
 throws_ok { $biotype1->so_acc('test') } qr/so_acc must be a Sequence Ontology accession/, 'so_acc() requires a SO acc like string';
 throws_ok { $biotype1->object_type('test') } qr/object_type must be gene or transcript/, 'object_type() must be gene or transcript';
 
 # test transcript biotype object
 my $transcript = $gene->canonical_transcript;
 debug("transcript biotype");
-is($transcript->biotype, 'protein_coding', "Trancript biotype is protein_coding");
+is($transcript->biotype, 'protein_coding', "Transcript biotype is protein_coding");
 my $biotype2 = $transcript->get_Biotype;
 ok($biotype2->isa("Bio::EnsEMBL::Biotype"), "Biotype object retrieved successfully");
 is($biotype2->object_type, 'transcript', 'Biotype is from Transcript object');
 is($biotype2->name, 'protein_coding', 'Biotype name is protein_coding');
 is($biotype2->biotype_group, 'coding', 'Biotype group is coding');
 is($biotype2->so_acc, 'SO:0000234', 'Biotype protein_coding refers to SO:0000234');
+is($biotype2->so_term, 'mRNA', 'Biotype protein_coding refers to SO term mRNA');
 ok($transcript->set_Biotype('new_biotype'), "Can successfully set new_biotype");
 throws_ok { $gene->set_Biotype() } qr/No argument provided/, 'set_Biotype() requires an argument';
 
@@ -77,6 +79,7 @@ is($biotype3->object_type, 'gene', 'Biotype is from Gene object');
 is($biotype3->name, 'tRNA', 'Biotype name is tRNA');
 is($biotype3->biotype_group, 'snoncoding', 'Biotype group is snoncoding');
 is($biotype3->so_acc, 'SO:0001263', 'Biotype tRNA refers to SO:0001263');
+is($biotype3->so_term, 'ncRNA_gene', 'Biotype protein_coding refers to SO term ncRNA_gene');
 
 # set biotype with term not in database
 debug("set biotype with term not in db");
@@ -87,6 +90,7 @@ is($biotype4->object_type, 'gene', 'Biotype is from Gene object');
 is($biotype4->name, 'dummy', 'Biotype name is dummy');
 is($biotype4->biotype_group, undef, 'Biotype group is not set');
 is($biotype4->so_acc, undef, 'Biotype SO acc is not set');
+is($biotype4->so_term, undef, 'Biotype SO term is not set');
 throws_ok { $gene->set_Biotype() } qr/No argument provided/, 'set_Biotype() requires an argument';
 
 # test fetch biotypes of object_type gene
@@ -95,7 +99,7 @@ my $biotypes1 = $biotype_adaptor->fetch_all_by_object_type('gene');
 is(ref $biotypes1, 'ARRAY', 'Got an array');
 is(scalar @{$biotypes1}, '2', 'of size 2');
 is_deeply($biotypes1, [$biotype1, $biotype3], 'with the correct objects');
-my $warning1 = warning { 
+my $warning1 = warning {
   $biotypes1 = $biotype_adaptor->fetch_all_by_object_type('none') };
 like( $warning1,
     qr/No objects retrieved. Check if object_type 'none' is correct./,

modules/t/cds.t

@@ -35,11 +35,10 @@ ok($db);
 
 my $stable_id = 'ENST00000217347';
 my $transcript_adaptor = $db->get_TranscriptAdaptor();
-my $transcript = 
-  $transcript_adaptor->fetch_by_stable_id($stable_id);
+my $transcript = $transcript_adaptor->fetch_by_stable_id($stable_id);
 
 
-my @cds = @{ $transcript->get_all_CDS() };  
+my @cds = @{ $transcript->get_all_CDS() };
 my @exons = @{ $transcript->get_all_translateable_Exons() };
 my $n = scalar(@cds);
 
@@ -50,6 +49,7 @@ for (my $i = 0; $i < $n; $i++) {
 
 is($cds[0]->start, $transcript->coding_region_start, "First cds is coding start");
 is($cds[$n-1]->end, $transcript->coding_region_end, "Last cds is coding end");
-is($cds[0]->feature_so_acc, 'SO:0000316', 'CDS feature SO acc is correct (CDS)');;
+is($cds[0]->feature_so_acc, 'SO:0000316', 'CDS feature SO acc is correct (CDS)');
+is($cds[0]->feature_so_term, 'CDS', 'CDS feature SO term is correct (CDS)');;
 
 done_testing();

modules/t/dnaDnaAlignFeature.t

@@ -159,5 +159,6 @@ ok($f);
 
 
 is($dnaf->feature_so_acc, 'SO:0000347', 'DnaDnaAlignFeature feature SO acc is correct (nucleotide_match)');
+is($dnaf->feature_so_term, 'nucleotide_match', 'DnaDnaAlignFeature feature SO term is correct (nucleotide_match)');
 
 done_testing();

modules/t/dnaPepAlignFeature.t

@@ -116,6 +116,7 @@ ok($dnaf->end == 16);
 ok( scalar($dnaf->ungapped_features) == 2);
 
 is($dnaf->feature_so_acc, 'SO:0000349', 'dnaPepAlignFeature feature SO acc is correct (protein_match)');
+is($dnaf->feature_so_term, 'protein_match', 'dnaPepAlignFeature feature SO term is correct (protein_match)');
 
 #
 # 12 Test retrieval from database