From b84e25a270a77338c020643568c395c237697255 Mon Sep 17 00:00:00 2001
From: Andrew Yates <ayates@ebi.ac.uk>
Date: Wed, 29 May 2013 14:45:22 +0000
Subject: [PATCH] [ENSCORESW-490]. Point raised by Dan. Default GFF3 for source
is . not ensembl. Using . and supporting basic features
---
modules/Bio/EnsEMBL/Utils/BiotypeMapper.pm | 3 +-
modules/Bio/EnsEMBL/Utils/IO/GFFSerializer.pm | 4 +-
modules/t/gffSerialiser.t | 41 +++++++++++++++++--
3 files changed, 41 insertions(+), 7 deletions(-)
diff --git a/modules/Bio/EnsEMBL/Utils/BiotypeMapper.pm b/modules/Bio/EnsEMBL/Utils/BiotypeMapper.pm
index 428b7fe62d..2a7fdd13d0 100644
--- a/modules/Bio/EnsEMBL/Utils/BiotypeMapper.pm
+++ b/modules/Bio/EnsEMBL/Utils/BiotypeMapper.pm
@@ -114,9 +114,10 @@ my %transcript_so_mapping = (
);
my %feature_so_mapping = (
+ 'Bio::EnsEMBL::Feature' => 'SO:0000001', # region
'Bio::EnsEMBL::Gene' => 'SO:0000704', # gene
'Bio::EnsEMBL::Transcript' => 'SO:0000673', # transcript
- 'Bio::EnsEMBL::Exon' => 'SO:0000147',
+ 'Bio::EnsEMBL::Exon' => 'SO:0000147', # exon
'Bio::EnsEMBL::Slice' => 'SO:0000001', # region
'Bio::EnsEMBL::SimpleFeature' => 'SO:0001411', # biological_region
'Bio::EnsEMBL::MiscFeature' => 'SO:0001411', # biological_region
diff --git a/modules/Bio/EnsEMBL/Utils/IO/GFFSerializer.pm b/modules/Bio/EnsEMBL/Utils/IO/GFFSerializer.pm
index 3274d2ba0b..d80846d297 100644
--- a/modules/Bio/EnsEMBL/Utils/IO/GFFSerializer.pm
+++ b/modules/Bio/EnsEMBL/Utils/IO/GFFSerializer.pm
@@ -52,7 +52,7 @@ my %strand_conversion = ( '1' => '+', '0' => '?', '-1' => '-');
Constructor
Arg [1] : Ontology Adaptor
Arg [2] : Optional File handle
- Arg [3] : Default source of the features. Defaults to ensembl
+ Arg [3] : Default source of the features. Defaults to .
Returntype : Bio::EnsEMBL::Utils::IO::GFFSerializer
@@ -77,7 +77,7 @@ sub new {
$self->{'stdout'} = 1;
}
if(!defined $self->{default_source}) {
- $self->{default_source} = 'ensembl';
+ $self->{default_source} = '.';
}
return $self;
}
diff --git a/modules/t/gffSerialiser.t b/modules/t/gffSerialiser.t
index 4fa0c8e557..f8a27ec23d 100644
--- a/modules/t/gffSerialiser.t
+++ b/modules/t/gffSerialiser.t
@@ -8,9 +8,10 @@ sub new {
return bless({}, ref($class) || $class);
}
+# default lookup and always returns region
sub name {
my ($self) = @_;
- return 'feature';
+ return 'region';
}
package Test::SO;
@@ -34,7 +35,10 @@ use warnings;
use Test::More;
use Bio::EnsEMBL::Test::MultiTestDB;
use Bio::EnsEMBL::Utils::IO::GFFSerializer;
+use Bio::EnsEMBL::Feature;
+use Bio::EnsEMBL::Slice;
use IO::String;
+use Test::Differences;
my $db = Bio::EnsEMBL::Test::MultiTestDB->new();
my $dba = $db->get_DBAdaptor('core');
@@ -53,7 +57,7 @@ my $ga = $dba->get_GeneAdaptor();
OUT
#Have to do this outside of the HERETO thanks to tabs
$expected .= join("\t",
- qw/20 ensembl feature 30274334 30300924 . + ./,
+ qw/20 ensembl region 30274334 30300924 . + ./,
'ID=ENSG00000131044;biotype=protein_coding;external_name=C20orf125;logic_name=ensembl'
);
$expected .= "\n";
@@ -61,6 +65,35 @@ OUT
assert_gff3($gene, $expected, 'Gene with no source serialises to GFF3 as expected. Source is ensembl');
}
+{
+ my $cs = $dba->get_CoordSystemAdaptor()->fetch_by_name('chromosome');
+ my $feature = Bio::EnsEMBL::Feature->new(
+ -SLICE => Bio::EnsEMBL::Slice->new(
+ -COORD_SYSTEM => $cs,
+ -SEQ => ('A'x10),
+ -SEQ_REGION_NAME => 'wibble',
+ -START => 1,
+ -END => 10
+ ),
+ -START => 1,
+ -END => 10,
+ -STRAND => 1,
+ );
+ my $expected = <<'OUT';
+##gff-version 3
+##sequence-region wibble 1 10
+OUT
+ #Have to do this outside of the HERETO thanks to tabs
+ $expected .= join("\t",
+ qw/wibble . region 1 10 . + ./,
+ ''
+ );
+ $expected .= "\n";
+
+ assert_gff3($feature, $expected, 'Default feature should seralise without attributes but leave a trailing \t');
+}
+
+
{
my $gene = $ga->fetch_by_stable_id($id);
$gene->source('wibble');
@@ -70,7 +103,7 @@ OUT
OUT
#Have to do this outside of the HERETO thanks to tabs
$expected .= join("\t",
- qw/20 wibble feature 30274334 30300924 . + ./,
+ qw/20 wibble region 30274334 30300924 . + ./,
'ID=ENSG00000131044;biotype=protein_coding;description=DJ310O13.1.2 (NOVEL PROTEIN SIMILAR DROSOPHILA PROTEIN CG7474%2C ISOFORM 2 ) (FRAGMENT). [Source:SPTREMBL%3BAcc:Q9BR18];external_name=C20orf125;logic_name=ensembl'
);
$expected .= "\n";
@@ -85,7 +118,7 @@ sub assert_gff3 {
my $ser = Bio::EnsEMBL::Utils::IO::GFFSerializer->new($ota, $fh);
$ser->print_main_header([$feature->feature_Slice()]);
$ser->print_feature($feature);
- is(${$fh->string_ref()}, $expected, $msg);
+ eq_or_diff(${$fh->string_ref()}, $expected, $msg);
}
done_testing();
\ No newline at end of file
--
GitLab