diff --git a/modules/Bio/EnsEMBL/DBSQL/ProteinFeatureAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/ProteinFeatureAdaptor.pm
index 87d804881f4f1115ae3f8a7b1881b9abb26918da..ebc4b469e5fbedeee00ce6172cf274237e5657ee 100755
--- a/modules/Bio/EnsEMBL/DBSQL/ProteinFeatureAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/ProteinFeatureAdaptor.pm
@@ -83,13 +83,13 @@ sub fetch_all_by_translation_id {
my @features;
my $analysis_adaptor = $self->db()->get_AnalysisAdaptor();
- my $sth = $self->prepare("SELECT protein_feature_id, p.seq_start, p.seq_end, p.analysis_id, " . " p.score, p.perc_ident, p.evalue, p.hit_start, p.hit_end, " . " p.hit_name, x.display_label, i.interpro_ac " . "FROM protein_feature p " . "LEFT JOIN interpro AS i ON p.hit_name = i.id " . "LEFT JOIN xref AS x ON x.dbprimary_acc = i.interpro_ac " . "WHERE p.translation_id = ?");
+ my $sth = $self->prepare("SELECT protein_feature_id, p.seq_start, p.seq_end, p.analysis_id, " . " p.score, p.perc_ident, p.evalue, p.hit_start, p.hit_end, " . " p.hit_name, p.hit_description, x.display_label, i.interpro_ac " . "FROM protein_feature p " . "LEFT JOIN interpro AS i ON p.hit_name = i.id " . "LEFT JOIN xref AS x ON x.dbprimary_acc = i.interpro_ac " . "WHERE p.translation_id = ?");
$sth->bind_param(1, $translation_id, SQL_INTEGER);
$sth->execute();
while (my $row = $sth->fetchrow_arrayref) {
- my ($dbID, $start, $end, $analysisid, $score, $perc_id, $evalue, $hstart, $hend, $hid, $desc, $interpro_ac) = @$row;
+ my ($dbID, $start, $end, $analysisid, $score, $perc_id, $evalue, $hstart, $hend, $hid, $hdesc, $desc, $interpro_ac) = @$row;
my $analysis = $analysis_adaptor->fetch_by_dbID($analysisid);
@@ -97,20 +97,21 @@ sub fetch_all_by_translation_id {
warning("Analysis with dbID=$analysisid does not exist\n" . "but is referenced by ProteinFeature $dbID");
}
- my $feat = Bio::EnsEMBL::ProteinFeature->new(-DBID => $dbID,
- -ADAPTOR => $self,
- -SEQNAME => $translation_id,
- -START => $start,
- -END => $end,
- -ANALYSIS => $analysis,
- -PERCENT_ID => $perc_id,
- -P_VALUE => $evalue,
- -SCORE => $score,
- -HSTART => $hstart,
- -HEND => $hend,
- -HSEQNAME => $hid,
- -IDESC => $desc,
- -INTERPRO_AC => $interpro_ac);
+ my $feat = Bio::EnsEMBL::ProteinFeature->new(-DBID => $dbID,
+ -ADAPTOR => $self,
+ -SEQNAME => $translation_id,
+ -START => $start,
+ -END => $end,
+ -ANALYSIS => $analysis,
+ -PERCENT_ID => $perc_id,
+ -P_VALUE => $evalue,
+ -SCORE => $score,
+ -HSTART => $hstart,
+ -HEND => $hend,
+ -HSEQNAME => $hid,
+ -HDESCRIPTION => $hdesc,
+ -IDESC => $desc,
+ -INTERPRO_AC => $interpro_ac);
push(@features, $feat);
} ## end while (my $row = $sth->fetchrow_arrayref)
@@ -205,23 +206,23 @@ sub store {
if (!defined($analysis)) {
throw("Feature doesn't have analysis. Can't write to database");
}
-
if (!$analysis->is_stored($db)) {
$db->get_AnalysisAdaptor->store($analysis);
}
- my $sth = $self->prepare("INSERT INTO protein_feature " . " SET translation_id = ?, " . " seq_start = ?, " . " seq_end = ?, " . " analysis_id = ?, " . " hit_start = ?, " . " hit_end = ?, " . " hit_name = ?, " . " score = ?, " . " perc_ident = ?, " . " evalue = ?");
+ my $sth = $self->prepare("INSERT INTO protein_feature " . " SET translation_id = ?, " . " seq_start = ?, " . " seq_end = ?, " . " analysis_id = ?, " . " hit_start = ?, " . " hit_end = ?, " . " hit_name = ?, " . " hit_description = ?, " . " score = ?, " . " perc_ident = ?, " . " evalue = ?");
- $sth->bind_param(1, $translation_id, SQL_INTEGER);
- $sth->bind_param(2, $feature->start, SQL_INTEGER);
- $sth->bind_param(3, $feature->end, SQL_INTEGER);
- $sth->bind_param(4, $analysis->dbID, SQL_INTEGER);
- $sth->bind_param(5, $feature->hstart, SQL_INTEGER);
- $sth->bind_param(6, $feature->hend, SQL_INTEGER);
- $sth->bind_param(7, $feature->hseqname, SQL_VARCHAR);
- $sth->bind_param(8, $feature->score, SQL_DOUBLE);
- $sth->bind_param(9, $feature->percent_id, SQL_FLOAT);
- $sth->bind_param(10, $feature->p_value, SQL_DOUBLE);
+ $sth->bind_param(1, $translation_id, SQL_INTEGER);
+ $sth->bind_param(2, $feature->start, SQL_INTEGER);
+ $sth->bind_param(3, $feature->end, SQL_INTEGER);
+ $sth->bind_param(4, $analysis->dbID, SQL_INTEGER);
+ $sth->bind_param(5, $feature->hstart, SQL_INTEGER);
+ $sth->bind_param(6, $feature->hend, SQL_INTEGER);
+ $sth->bind_param(7, $feature->hseqname, SQL_VARCHAR);
+ $sth->bind_param(8, $feature->hdescription, SQL_LONGVARCHAR);
+ $sth->bind_param(9, $feature->score, SQL_DOUBLE);
+ $sth->bind_param(10, $feature->percent_id, SQL_FLOAT);
+ $sth->bind_param(11, $feature->p_value, SQL_DOUBLE);
$sth->execute();
@@ -274,7 +275,7 @@ sub save {
my $db = $self->db();
my $analysis_adaptor = $db->get_AnalysisAdaptor();
- my $sql = qq{INSERT INTO $tablename (translation_id, seq_start, seq_end, hit_start, hit_end, hit_name, analysis_id, score, evalue, perc_ident, external_data) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)};
+ my $sql = qq{INSERT INTO $tablename (translation_id, seq_start, seq_end, hit_start, hit_end, hit_name, hdescription, analysis_id, score, evalue, perc_ident, external_data) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)};
my $sth = $self->prepare($sql);
@@ -309,11 +310,12 @@ sub save {
$sth->bind_param(4, $hstart, SQL_INTEGER);
$sth->bind_param(5, $hend, SQL_INTEGER);
$sth->bind_param(6, $feat->hseqname, SQL_VARCHAR);
- $sth->bind_param(7, $feat->analysis->dbID, SQL_INTEGER);
- $sth->bind_param(8, $feat->score, SQL_DOUBLE);
- $sth->bind_param(9, $feat->p_value, SQL_DOUBLE);
- $sth->bind_param(10, $feat->percent_id, SQL_FLOAT);
- $sth->bind_param(11, $extra_data, SQL_LONGVARCHAR);
+ $sth->bind_param(7, $feat->hdescription, SQL_LONGVARCHAR);
+ $sth->bind_param(8, $feat->analysis->dbID, SQL_INTEGER);
+ $sth->bind_param(9, $feat->score, SQL_DOUBLE);
+ $sth->bind_param(10, $feat->p_value, SQL_DOUBLE);
+ $sth->bind_param(11, $feat->percent_id, SQL_FLOAT);
+ $sth->bind_param(12, $extra_data, SQL_LONGVARCHAR);
$sth->execute();
$original->dbID($sth->{'mysql_insertid'});
diff --git a/modules/Bio/EnsEMBL/FeaturePair.pm b/modules/Bio/EnsEMBL/FeaturePair.pm
index 5cbcbf1d5a7bff1340f1021381606a4c14ede648..170d504cd0c48a40ae63275423f765c9a6aaaa5d 100755
--- a/modules/Bio/EnsEMBL/FeaturePair.pm
+++ b/modules/Bio/EnsEMBL/FeaturePair.pm
@@ -10,6 +10,7 @@
http://www.ensembl.org/info/about/code_licence.html
=head1 CONTACT
+
Please email comments or questions to the public Ensembl
developers list at <dev@ensembl.org>.
@@ -126,7 +127,7 @@ sub new {
my $self = $class->SUPER::new(@_);
- my ($hstart, $hend, $hstrand, $percent_id, $score, $species, $hspecies, $p_value, $hseqname, $f1, $f2, $coverage, $hcoverage, $group_id, $level_id, $external_db_id, $extra_data, $external_db_name, $external_display_db_name) = rearrange(['HSTART', 'HEND', 'HSTRAND', 'PERCENT_ID', 'SCORE', 'SPECIES', 'HSPECIES', 'P_VALUE', 'HSEQNAME', 'FEATURE1', 'FEATURE2', 'COVERAGE', 'HCOVERAGE', 'GROUP_ID', 'LEVEL_ID', 'EXTERNAL_DB_ID', 'EXTRA_DATA', 'DBNAME', 'DB_DISPLAY_NAME'], @_);
+ my ($hstart, $hend, $hstrand, $percent_id, $score, $species, $hspecies, $p_value, $hseqname, $f1, $f2, $coverage, $hcoverage, $group_id, $level_id, $external_db_id, $extra_data, $external_db_name, $external_display_db_name, $hdescription) = rearrange(['HSTART', 'HEND', 'HSTRAND', 'PERCENT_ID', 'SCORE', 'SPECIES', 'HSPECIES', 'P_VALUE', 'HSEQNAME', 'FEATURE1', 'FEATURE2', 'COVERAGE', 'HCOVERAGE', 'GROUP_ID', 'LEVEL_ID', 'EXTERNAL_DB_ID', 'EXTRA_DATA', 'DBNAME', 'DB_DISPLAY_NAME', 'HDESCRIPTION'], @_);
if (defined($hstart) && defined($hend) && ($hend < $hstart)) {
throw('HSTART must be less than or equal to HEND');
@@ -153,7 +154,7 @@ sub new {
$self->{'extra_data'} = $extra_data;
$self->{'dbname'} = $external_db_name;
$self->{'db_display_name'} = $external_display_db_name;
-
+ $self->{'hdescription'} = $hdescription;
#
# Feature1 and Feature2 arg handling for backwards compatibility
#
@@ -583,6 +584,24 @@ sub p_value {
return $self->{'p_value'};
}
+=head2 hdescription
+
+ Arg [1] : String (optional)
+ Example : $des = $fp->hdescription()
+ Description: Getter Setter for optional description of this feature
+ Returntype : String
+ Exceptions : none
+ Caller : general
+ Status : Stable
+
+=cut
+
+sub hdescription {
+ my $self = shift;
+ $self->{'hdescription'} = shift if (@_);
+ return $self->{'hdescription'};
+}
+
=head2 display_id
Arg [1] : none
@@ -768,11 +787,12 @@ sub invert {
$slice = $self->hslice;
}
- my $hstart = $self->{'hstart'};
- my $hend = $self->{'hend'};
- my $hstrand = $self->{'hstrand'};
- my $hspecies = $self->{'hspecies'};
- my $hseqname = $self->{'hseqname'};
+ my $hstart = $self->{'hstart'};
+ my $hend = $self->{'hend'};
+ my $hstrand = $self->{'hstrand'};
+ my $hspecies = $self->{'hspecies'};
+ my $hseqname = $self->{'hseqname'};
+ my $hdescription = $self->{'hdescription'};
my $start = $self->{'start'};
my $end = $self->{'end'};
@@ -792,6 +812,8 @@ sub invert {
$self->{'hseqname'} = $seqname;
$self->{'hspecies'} = $species;
+ $self->{'hdescription'} = $hdescription;
+
$self->{'hslice'} = $self->slice;
$self->{'slice'} = $slice;
} ## end sub invert
diff --git a/modules/t/proteinFeature.t b/modules/t/proteinFeature.t
index 538533f511279e21fb7c5c072d4a1e46b4c7c457..576cd597bcf35ca17df75d7d617d42ce2491ce02 100644
--- a/modules/t/proteinFeature.t
+++ b/modules/t/proteinFeature.t
@@ -1,10 +1,10 @@
use strict;
use warnings;
-BEGIN {
- $| = 1;
- use Test;
- plan tests => 17;
+
+BEGIN { $| = 1;
+ use Test;
+ plan tests => 17;
}
use Bio::EnsEMBL::Test::TestUtils;
@@ -12,41 +12,48 @@ use Bio::EnsEMBL::Test::TestUtils;
use Bio::EnsEMBL::ProteinFeature;
use Bio::EnsEMBL::Analysis;
-our $verbose = 0; #turn on or off debug statements
+our $verbose = 0; #turn on or off debug statements
+
#
# Test new and getters
#
-my $start = 10;
-my $end = 100;
-my $hstart = 1;
-my $hend = 90;
-my $hstrand = 1;
-my $hseqname = 'RF1231';
+my $start = 10;
+my $end = 100;
+my $hstart = 1;
+my $hend = 90;
+my $hstrand = 1;
+my $hseqname = 'RF1231';
my $percent_id = 90.8;
-my $p_value = '1.52';
-my $score = 50;
-my $species = 'Homo_sapiens';
-my $hspecies = 'Mus_musculus';
+my $p_value = '1.52';
+my $score = 50;
+my $species = 'Homo_sapiens';
+my $hspecies = 'Mus_musculus';
+my $hdes = "Hit description";
-my $idesc = 'interpro description';
+my $idesc = 'interpro description';
my $interpro_ac = 'interpro accession';
my $analysis = Bio::EnsEMBL::Analysis->new(-LOGIC_NAME => 'test');
-my $f = Bio::EnsEMBL::ProteinFeature->new(-START => $start,
- -END => $end,
- -ANALYSIS => $analysis,
- -HSTART => $hstart,
- -HEND => $hend,
- -HSEQNAME => $hseqname,
- -PERCENT_ID => $percent_id,
- -P_VALUE => $p_value,
- -SCORE => $score,
- -SPECIES => $species,
- -HSPECIES => $hspecies,
- -IDESC => $idesc,
- -INTERPRO_AC => $interpro_ac);
+
+my $f = Bio::EnsEMBL::ProteinFeature->new
+ (-START => $start,
+ -END => $end,
+ -ANALYSIS => $analysis,
+ -HSTART => $hstart,
+ -HEND => $hend,
+ -HSEQNAME => $hseqname,
+ -PERCENT_ID => $percent_id,
+ -P_VALUE => $p_value,
+ -SCORE => $score,
+ -SPECIES => $species,
+ -HSPECIES => $hspecies,
+ -HDESCRIPTION=> $hdes,
+ -IDESC => $idesc,
+ -INTERPRO_AC => $interpro_ac);
+
+
ok($f && $f->isa('Bio::EnsEMBL::ProteinFeature'));
@@ -55,15 +62,16 @@ ok($f->end == $end);
ok($f->analysis == $analysis);
ok($f->hstart == $hstart);
-ok($f->hend == $hend);
+ok($f->hend == $hend);
ok($f->hseqname eq $hseqname);
ok($f->percent_id == $percent_id);
ok($f->p_value == $p_value);
ok($f->score == $score);
-ok($f->species eq $species);
+ok($f->species eq $species);
ok($f->hspecies eq $hspecies);
+ok($f->hdescription eq $hdes);
-ok($f->idesc eq $idesc);
+ok($f->idesc eq $idesc);
ok($f->interpro_ac eq $interpro_ac);
# check that the strand is 0
@@ -72,6 +80,9 @@ ok($f->strand == 0);
#
# Test setters
#
-ok(test_getter_setter($f, 'idesc', 'interpro desc1'));
-ok(test_getter_setter($f, 'interpro_ac', 'interpro ac1'));
+ok(test_getter_setter($f, 'idesc', 'interpro desc1'));
+ok(test_getter_setter($f, 'interpro_ac', 'interpro ac1'));
+
+
+
diff --git a/modules/t/proteinFeatureAdaptor.t b/modules/t/proteinFeatureAdaptor.t
index ed3b48391ef02600f210a4d9b298083b64d0334d..081e4f2c980cfc2884d66ab4a6c12c37bfa94554 100644
--- a/modules/t/proteinFeatureAdaptor.t
+++ b/modules/t/proteinFeatureAdaptor.t
@@ -38,3 +38,49 @@ sub print_features {
}
}
}
+
+# test adding and retrieving a new feature
+my $start = 10;
+my $end = 100;
+my $hstart = 1;
+my $hend = 90;
+my $hstrand = 1;
+my $hseqname = 'RF1231';
+my $percent_id = 90.8;
+my $p_value = '1.52';
+my $score = 50;
+my $species = 'Homo_sapiens';
+my $hspecies = 'Mus_musculus';
+my $hdes = "Hit description";
+
+my $idesc = 'interpro description';
+my $interpro_ac = 'interpro accession';
+
+my $analysis = Bio::EnsEMBL::Analysis->new(-LOGIC_NAME => 'test');
+
+
+my $f = Bio::EnsEMBL::ProteinFeature->new
+ (-START => $start,
+ -END => $end,
+ -ANALYSIS => $analysis,
+ -HSTART => $hstart,
+ -HEND => $hend,
+ -HSEQNAME => $hseqname,
+ -PERCENT_ID => $percent_id,
+ -P_VALUE => $p_value,
+ -SCORE => $score,
+ -SPECIES => $species,
+ -HSPECIES => $hspecies,
+ -HDESCRIPTION=> $hdes,
+ -IDESC => $idesc,
+ -INTERPRO_AC => $interpro_ac);
+
+$pfa->store($f,21724);
+
+my $pfs = $pfa->fetch_all_by_translation_id(21724);
+
+ok(@$pfs == 16);
+
+my @pfs = grep{$_->hdescription() eq $hdes} @$pfs;
+
+ok(scalar @pfs > 0);
diff --git a/sql/patch_69_70_e.sql b/sql/patch_69_70_e.sql
new file mode 100644
index 0000000000000000000000000000000000000000..b11061a1587dd67d3bef8b16871a79fea0a95db5
--- /dev/null
+++ b/sql/patch_69_70_e.sql
@@ -0,0 +1,13 @@
+# patch_69_70_e.sql
+#
+# Title: Add hit_description to protein_feature
+#
+# Description: Column allowing an optional description to be added to a protein_feature
+
+ALTER TABLE protein_feature ADD COLUMN hit_description TEXT;
+
+# Patch identifier
+INSERT INTO meta (species_id, meta_key, meta_value)
+ VALUES (NULL, 'patch', 'patch_69_70_e.sql|protein_feature_hit_description');
+
+
diff --git a/sql/table.sql b/sql/table.sql
index a865a4744705e0f8ebacd9b64516a3b313ccb5cf..2d7192e94d3a0b57fc0122c937d3e8d36f92a7c4 100755
--- a/sql/table.sql
+++ b/sql/table.sql
@@ -1594,6 +1594,7 @@ CREATE TABLE protein_align_feature (
@column hit_start Alignment hit start position.
@column hit_end Alignment hit end position.
@column hit_name Alignment hit name.
+@column hit_description Optional description of the hit.
@column analysis_id Foreign key references to the @link analysis table.
@column score Alignment score.
@column evalue Alignment E-value.
@@ -1614,6 +1615,7 @@ CREATE TABLE protein_feature (
hit_start INT(10) NOT NULL,
hit_end INT(10) NOT NULL,
hit_name VARCHAR(40) NOT NULL,
+ hit_description TEXT,
analysis_id SMALLINT UNSIGNED NOT NULL,
score DOUBLE,
evalue DOUBLE,