diff --git a/modules/Bio/EnsEMBL/Gene.pm b/modules/Bio/EnsEMBL/Gene.pm
index 841eef206b50af738415c8cfaac7ac266e6a3b0f..265507cb335e5ecbd211cb4dfb209d8a0b02bbc0 100755
--- a/modules/Bio/EnsEMBL/Gene.pm
+++ b/modules/Bio/EnsEMBL/Gene.pm
@@ -380,12 +380,14 @@ sub dbID {
Arg [1] : string $external_name
Example : none
- Description: get/set for attribute external_name. It could be calculated
- from dblinks in a species dependent way. Well introduce
- that later.
+ Description: get/set for attribute external_name. It initially calculates
+ the longest transcript for the gene in question and then
+ delegates the call to the external_name method on Transcript.
+ Species dependant searching is handled by this method on
+ Transcript.
Returntype : string
Exceptions : none
- Caller : Lite::GeneAdaptor knows how to set it correct
+ Caller : general
=cut
@@ -395,36 +397,92 @@ sub external_name {
if( defined $arg ) {
$self->{'_external_name'} = $arg;
}
+ else {
+ # find the transcript with the longest length which is
+ # attached to this gene. Use the longest length as the
+ # determining factor when having to select between transcripts
+ # and their external references
+ my $transcript = $self->_get_longest_Transcript;
+
+ $self->{'_external_name'} = $transcript->external_name;
+ }
- return $self->{'_external_name'};
+ return $self->{'_external_name'};
}
-
=head2 external_db
Arg [1] : string $external_db
Example : none
Description: get/set for attribute external_db. The db is the one that
- belongs to the external_name
+ belongs to the external_name. It initially calculates
+ the longest transcript for the gene in question and then
+ delegates the call to the external_db method on Transcript.
+ Species dependant searching is handled by this method on
+ Transcript.
Returntype : string
Exceptions : none
Caller : general
=cut
-
sub external_db {
my ($self, $arg ) = @_;
if( defined $arg ) {
$self->{'_external_db'} = $arg;
}
+ else {
+ # find the transcript with the longest length which is
+ # attached to this gene. Use the longest length as the
+ # determining factor when having to select between transcripts
+ # and their external references
+ my $transcript = $self->_get_longest_Transcript;
+
+ $self->{'_external_db'} = $transcript->external_db;
+ }
return $self->{'_external_db'};
}
+=head2 _get_longest_Transcript
+
+ Args : none
+ Example : none
+ Description: An INTERNAL method which determines the longest transcript
+ for the given gene.get/set for attribute external_db. The db is the one that
+ belongs to the external_name. It initially calculates
+ the longest transcript for the gene in question and then
+ delegates the call to the external_db method on Transcript.
+ Species dependant searching is handled by this method on
+ Transcript.
+ Returntype : a single Bio::EnsEMBL::Transcript
+ Exceptions : none
+ Caller : external_name and external_db methods on Gene.pm
+
+=cut
+
+sub _get_longest_Transcript {
+ my $self = shift;
+
+ my $transcripts = $self->get_all_Transcripts;
+
+ my $longest_index = 0;
+ my $longest_length = 0;
+ my $tran_count = 0;
+
+ foreach my $trans ( @{$transcripts} ) {
+ if ( $trans->length > $longest_length ) {
+ $longest_length = $trans->length;
+ $longest_index = $tran_count;
+ }
+ $tran_count++;
+ }
+
+ return $transcripts->[$longest_index];
+}
=head2 description
@@ -883,7 +941,7 @@ sub temporary_id {
Description: You can set the species for this gene if you want to use species
specific behaviour. Otherwise species is retrieved from attached
database.
- Returntype : Bio::EnsEMBL::Species
+ Returntype : Bio::Species
Exceptions : none
Caller : external_name, external_db, general for setting
diff --git a/modules/Bio/EnsEMBL/Transcript.pm b/modules/Bio/EnsEMBL/Transcript.pm
index 478bcb6c1284683ac7cdca4041c4f3d70c3c443a..df4ff6b1d42307b39d7f070a4646cb8aaba0a0dc 100755
--- a/modules/Bio/EnsEMBL/Transcript.pm
+++ b/modules/Bio/EnsEMBL/Transcript.pm
@@ -167,6 +167,9 @@ sub external_db {
if( defined $arg ) {
$self->{'_external_db'} = $arg;
}
+ else {
+ $self->{'_external_db'} = $self->_get_external_info("db");
+ }
return $self->{'_external_db'};
}
@@ -184,16 +187,108 @@ sub external_db {
=cut
sub external_name {
- my ($self, $arg ) = @_;
+ my ($self, $arg) = @_;
if( defined $arg ) {
$self->{'_external_name'} = $arg;
}
+ else {
+ $self->{'_external_name'} = $self->_get_external_info("name");
+ }
return $self->{'_external_name'};
}
+=head2 _get_external_info
+
+ Title : _get_external_info
+ Usage : $ext_name = $obj->_get_external_info();
+ Function: external_name if available
+ Example :
+ Returns : the external name of this transcript
+ Args : string. Switch on whether to return a name or dbname.
+
+=cut
+
+sub _get_external_info {
+ my ($self, $required) = @_;
+
+ # find out from which species this translation comes from
+ my $species = $self->species->species;
+
+ # go and grab the list of DBLinks
+ my $dblinks = $self->get_all_DBLinks;
+
+ # set the priority of the order in which the external dbs are searched
+ # based on the species
+ # the actual order of dbs was determined by the deprecated priority column
+ # in the external_db table
+
+ my @priority_order = [];
+
+ # the kind of case statment switching is performed on the first records
+ # from the meta table of the relevant species.
+
+ # human
+ if ( $species eq 'sapiens' ) {
+ @priority_order = qw{ HUGO SWISSPROT SPTREMBL RefSeq LocusLink };
+ }
+ # anopheles
+ elsif ( $species eq 'gambiae' ) {
+ @priority_order = qw{ ANOSUB SWISSPROT SPTREMBL };
+ }
+ # zebra fish
+ elsif ( $species eq 'rerio' ) {
+ @priority_order = qw{ SWISSPROT SPTREMBL };
+ }
+ # fugu
+ elsif ( $species eq 'rubripes' ) {
+ @priority_order = qw{ SWISSPROT SPTREMBL RefSeq LocusLink HUGO };
+ }
+ # mouse
+ elsif ( $species eq 'musculus' ) {
+ @priority_order = qw{ MarkerSymbol SWISSPROT RefSeq LocusLink };
+ }
+ # default list if species is not set
+ else {
+ $self->warn("Transcript::external_name - No species set. Using default DB order.");
+ @priority_order = qw{ HUGO SWISSPROT SPTREMBL RefSeq LocusLink };
+ }
+
+ # find a match (first one) for the db with the highest available priority
+ my $name = undef;
+ my $db = undef;
+
+ # we would hope that each transcript has only a single DBLink per db but
+ # implement as a loop just in case, taking the first relevant record found
+ foreach my $curr_db ( @priority_order ) {
+ foreach my $dbl ( @{$dblinks} ) {
+ if ( $curr_db eq $dbl->dbname ) {
+ $name = $dbl->primary_id;
+ $db = $dbl->dbname;
+ last;
+ }
+ }
+ if ( defined $name ) {
+ last;
+ }
+ }
+
+ if ( $required eq 'name' ) {
+ return $name;
+ }
+ elsif ( $required eq 'db' ) {
+ return $db;
+ }
+ else {
+ $self->warn("Transcript::_get_external_info - no xref data could be retrieved.");
+ return undef;
+ }
+
+}
+
+
sub is_known {
my $self = shift;
if( defined $self->external_name() && $self->external_name() ne '' ) {
@@ -1467,6 +1562,36 @@ sub species {
+=head2 species
+
+ Arg [1] : optional Bio::Species $species
+ Example : none
+ Description: You can set the species for this gene if you want to use species
+ specific behaviour. Otherwise species is retrieved from attached
+ database.
+ Returntype : Bio::Species
+ Exceptions : none
+ Caller : external_name, external_db, general for setting
+
+=cut
+
+
+sub species {
+ my ( $self, $species ) = @_;
+
+ if( defined $species ) {
+ $self->{species} = $species;
+ } else {
+ if( ! exists $self->{species} ) {
+ if( defined $self->adaptor() ) {
+ $self->{species} = $self->adaptor()->db->get_MetaContainer()
+ ->get_Species();
+ }
+ }
+ }
+
+ return $self->{species};
+}
##########################################################