From 24c8a5fbe62bf4a58968dbbb9639dd983ba31972 Mon Sep 17 00:00:00 2001
From: Kieron Taylor <ktaylor@ebi.ac.uk>
Date: Thu, 7 Feb 2013 13:06:16 +0000
Subject: [PATCH] Introduced fetching by stable ID lists
---
modules/Bio/EnsEMBL/DBSQL/BaseAdaptor.pm | 63 ++++++++++++++++---
.../Bio/EnsEMBL/DBSQL/BaseFeatureAdaptor.pm | 27 +++++++-
modules/t/baseAdaptor.t | 11 ++++
3 files changed, 92 insertions(+), 9 deletions(-)
diff --git a/modules/Bio/EnsEMBL/DBSQL/BaseAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/BaseAdaptor.pm
index 9190dc45cc..ffa4533c33 100755
--- a/modules/Bio/EnsEMBL/DBSQL/BaseAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/BaseAdaptor.pm
@@ -92,7 +92,7 @@ use vars qw(@ISA @EXPORT);
use strict;
use Bio::EnsEMBL::Utils::Exception qw(throw);
-use Bio::EnsEMBL::Utils::Scalar qw(assert_ref);
+use Bio::EnsEMBL::Utils::Scalar qw(assert_ref assert_integer);
use DBI qw(:sql_types);
use Data::Dumper;
@@ -654,6 +654,34 @@ sub fetch_all_by_dbID_list {
sub _uncached_fetch_all_by_dbID_list {
my ( $self, $id_list_ref, $slice ) = @_;
+ return $self->_uncached_fetch_all_by_id_list($id_list_ref, $slice, "dbID");
+} ## end sub fetch_all_by_dbID_list
+
+=head2 _uncached_fetch_all_by_id_list
+
+ Arg [1] : listref of IDs
+ Arg [2] : (optional) Bio::EnsEMBL::Slice $slice
+ A slice that features should be remapped to
+ Arg [3] : String describing the ID type.
+ Valid values include dbID and stable_id. dbID is an alias for
+ the primary key, while other names map directly to table columns
+ of the Feature this adaptor manages.
+ Example : $list_of_features = $adaptor->_uncached_fetch_all_by_id_list(
+ [qw(ENSG00000101321 ENSG00000101346 ENSG00000101367)],
+ undef,
+ "stable_id");
+ Description: This is a generic method used to fetch lists of features by IDs.
+ It avoids caches, meaning it is best suited for block fetching.
+ See fetch_all_by_dbID_list() for more info.
+ Returntype : ListRef of Bio::EnsEMBL::Feature
+ Exceptions : Thrown if a list of IDs is not supplied.
+ Caller : BaseFeatureAdaptor, BaseAdaptor and derived classes.
+
+=cut
+
+sub _uncached_fetch_all_by_id_list {
+ my ( $self, $id_list_ref, $slice, $id_type ) = @_;
+
if ( !defined($id_list_ref) || ref($id_list_ref) ne 'ARRAY' ) {
throw("id_list list reference argument is required");
}
@@ -670,9 +698,22 @@ sub _uncached_fetch_all_by_dbID_list {
# length of 16, this means 2048 dbIDs in each query.
my $max_size = 2048;
-
+ # prepare column name for query
+ my $field_name;
+ if ($id_type eq "dbID") {
+ $field_name = $name."_id";
+ } else {
+ $field_name = $id_type;
+ }
+
+ # build up unique id list, also validate on the way by
my %id_list;
- $id_list{$_}++ for @{$id_list_ref};
+ for (@{$id_list_ref}) {
+ $id_list{$_}++;
+ if ($id_type ne "stable_id") {
+ assert_integer($_,"$field_name");
+ }
+ }
my @id_list = keys %id_list;
my @out;
@@ -687,20 +728,26 @@ sub _uncached_fetch_all_by_dbID_list {
@ids = @id_list;
@id_list = ();
}
-
+
if ( scalar(@ids) > 1 ) {
- $id_str = " IN (" . join( ',', @ids ) . ")";
+ # stable_ids are the only feature attribute which is expressed as a
+ # varchar. These need to be quoted or the SQL will bounce
+ if ($id_type eq "stable_id") {
+ $id_str = " IN (" . join( ',', map qq("$_"), @ids ) . ")";
+ } else {
+ $id_str = " IN (" . join( ',', @ids). ")";
+ }
} else {
$id_str = " = " . $ids[0];
}
-
- my $constraint = "${syn}.${name}_id $id_str";
+
+ my $constraint = "${syn}.${field_name} $id_str";
push @out, @{ $self->generic_fetch($constraint, undef, $slice) };
}
return \@out;
-} ## end sub fetch_all_by_dbID_list
+}
# might not be a good idea, but for convenience
# shouldnt be called on the BIG tables though
diff --git a/modules/Bio/EnsEMBL/DBSQL/BaseFeatureAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/BaseFeatureAdaptor.pm
index c4d271ffdd..1ecb522c89 100644
--- a/modules/Bio/EnsEMBL/DBSQL/BaseFeatureAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/BaseFeatureAdaptor.pm
@@ -504,7 +504,7 @@ sub fetch_all_by_Slice_constraint {
=head2 fetch_all_by_logic_name
- Arg [3] : string $logic_name
+ Arg [1] : string $logic_name
the logic name of the type of features to obtain
Example : $fs = $a->fetch_all_by_logic_name('foobar');
Description: Returns a listref of features created from the database.
@@ -536,6 +536,31 @@ sub fetch_all_by_logic_name {
return $self->generic_fetch($constraint);
}
+=head2 fetch_all_by_stable_id_list
+
+ Arg [1] : string $logic_name
+ the logic name of the type of features to obtain
+ Arg [2] : Bio::EnsEMBL::Slice $slice
+ the slice from which to obtain features
+ Example : $fs = $a->fetch_all_by_stable_id_list(["ENSG00001","ENSG00002", ...]);
+ Description: Returns a listref of features identified by their stable IDs.
+ This method only fetches features of the same type as the calling
+ adaptor.
+ Results are constrained to a slice if the slice is provided.
+ Returntype : listref of Bio::EnsEMBL::Feature
+ Exceptions : thrown if no stable ID list is provided.
+ Caller : General
+ Status : Stable
+
+=cut
+
+# Adapted from BaseAdaptor->uncached_fetch_all_by_dbID_list
+sub fetch_all_by_stable_id_list {
+ my ( $self, $id_list_ref, $slice ) = @_;
+
+ return $self->_uncached_fetch_all_by_id_list($id_list_ref,$slice,"stable_id");
+}
+
# Method that creates an object. Called by the _objs_from_sth() method
# in the sub-classes (the various feature adaptors). Overridden by the
# feature collection classes.
diff --git a/modules/t/baseAdaptor.t b/modules/t/baseAdaptor.t
index 5d7041b9fc..f8229211ab 100644
--- a/modules/t/baseAdaptor.t
+++ b/modules/t/baseAdaptor.t
@@ -1,6 +1,7 @@
use strict;
use warnings;
use Test::More;
+use Test::Exception;
use Bio::EnsEMBL::Test::MultiTestDB;
use DBI qw/:sql_types/;
@@ -12,4 +13,14 @@ $gene_adaptor->bind_param_generic_fetch('protein_coding', SQL_VARCHAR);
my $count = $gene_adaptor->generic_count('g.biotype =?');
is($count, 20, 'Checking generic_count for protein_coding genes returns expected amounts');
+
+# fetch_all_by_dbID_list tests
+
+my $gene_list = $gene_adaptor->_uncached_fetch_all_by_id_list([qw(ENSG00000101321 ENSG00000101346 ENSG00000101367)],undef,"stable_id");
+ok(scalar(@$gene_list) == 3, "Basic uncached fetch by list");
+
+dies_ok {
+ $gene_list = $gene_adaptor->_uncached_fetch_all_by_id_list([qw(ENSG00000101321 ENSG00000101346 ENSG00000101367)],undef,"dbID")
+} "Wrong data type for ID";
+
done_testing();
\ No newline at end of file
--
GitLab