Merge pull request #272 from Ensembl/feature/pipeline_xref_datacheck

Create new datacheck groups for xref-related pipelines.
Ensembl · Jul 10, 2020 · 291776e · 291776e
2 parents 8326bae + ea3cc84
commit 291776e
Show file tree

Hide file tree

Showing 8 changed files with 31 additions and 17 deletions.
diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/CompareGOXref.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/CompareGOXref.pm
@@ -30,7 +30,7 @@ extends 'Bio::EnsEMBL::DataCheck::DbCheck';
 use constant {
   NAME           => 'CompareGOXref',
   DESCRIPTION    => 'Compare GO xref counts between two databases, categorised by source',
-  GROUPS         => ['compare_core', 'xref'],
+  GROUPS         => ['compare_core', 'xref', 'xref_go_projection'],
   DATACHECK_TYPE => 'advisory',
   DB_TYPES       => ['core']
 };

diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/CompareProjectedGOXrefs.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/CompareProjectedGOXrefs.pm
@@ -30,7 +30,7 @@ extends 'Bio::EnsEMBL::DataCheck::DbCheck';
 use constant {
   NAME           => 'CompareProjectedGOXrefs',
   DESCRIPTION    => 'Compare GO xref counts between two databases, categorised by source coming from the info_type',
-  GROUPS         => ['compare_core', 'xref'],
+  GROUPS         => ['compare_core', 'xref', 'xref_go_projection'],
   DATACHECK_TYPE => 'advisory',
   DB_TYPES       => ['core'],
   TABLES         => ['xref']
@@ -52,7 +52,7 @@ sub go_xref_counts {
   my ($self, $old_dba) = @_;
 
   my $minimum_count = 500;
-  my $threshold = 0.80;
+  my $threshold = 0.66;
 
   my $desc = "Consistent GO xref counts between ".
              $self->dba->dbc->dbname.

diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/CompareProjectedGeneNames.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/CompareProjectedGeneNames.pm
@@ -30,7 +30,7 @@ extends 'Bio::EnsEMBL::DataCheck::DbCheck';
 use constant {
   NAME           => 'CompareProjectedGeneNames',
   DESCRIPTION    => 'Compare Projected Gene Name counts between two databases',
-  GROUPS         => ['compare_core', 'xref'],
+  GROUPS         => ['compare_core', 'xref', 'xref_name_projection'],
   DATACHECK_TYPE => 'advisory',
   DB_TYPES       => ['core'],
   TABLES         => ['xref','gene','object_xref','seq_region','coord_system']
@@ -51,7 +51,7 @@ sub tests {
 sub projected_gene_name_counts {
   my ($self, $old_dba) = @_;
 
-  my $threshold = 0.80;
+  my $threshold = 0.66;
 
   my $desc = "Checking Projected Gene Names between ".
              $self->dba->dbc->dbname.

diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/CompareProjectedSynonyms.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/CompareProjectedSynonyms.pm
@@ -30,7 +30,7 @@ extends 'Bio::EnsEMBL::DataCheck::DbCheck';
 use constant {
   NAME           => 'CompareProjectedSynonyms',
   DESCRIPTION    => 'Compare Projected Synonyms counts between two databases, categorised by db_name coming from the external_db',
-  GROUPS         => ['compare_core', 'xref'],
+  GROUPS         => ['compare_core', 'xref', 'xref_name_projection'],
   DATACHECK_TYPE => 'advisory',
   DB_TYPES       => ['core'],
   TABLES         => ['xref','external_db','external_synonym','object_xref']
@@ -51,7 +51,7 @@ sub tests {
 sub projected_synonyms_counts {
   my ($self, $old_dba) = @_;
 
-  my $threshold = 0.80;
+  my $threshold = 0.66;
 
   my $desc = "Checking Projected Synonyms between ".
              $self->dba->dbc->dbname.
@@ -77,4 +77,4 @@ sub projected_synonyms_counts {
   my $sql2 = sprintf($sql, $old_dba->species_id);
   row_subtotals($self->dba, $old_dba, $sql1, $sql2, $threshold, $desc);
 }
-1;
+1;
diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/CompareXref.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/CompareXref.pm
@@ -83,7 +83,7 @@ sub xref_counts {
   /;
   my $sql1 = sprintf($sql, $self->dba->species_id);
   my $sql2 = sprintf($sql, $old_dba->species_id);
-  row_subtotals($self->dba, $old_dba, $sql1, $sql2, 0.70, $desc);
+  row_subtotals($self->dba, $old_dba, $sql1, $sql2, 0.66, $desc);
 }
 
 1;
diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/DisplayXrefExists.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/DisplayXrefExists.pm
@@ -30,11 +30,19 @@ extends 'Bio::EnsEMBL::DataCheck::DbCheck';
 use constant {
   NAME           => 'DisplayXrefExists',
   DESCRIPTION    => 'At least one gene name exists',
-  GROUPS         => ['core', 'xref'],
+  GROUPS         => ['core', 'xref', 'xref_name_projection'],
   DATACHECK_TYPE => 'advisory',
   TABLES         => ['coord_system', 'gene', 'seq_region', 'transcript', 'xref'],
 };
 
+sub skip_tests {
+  my ($self) = @_;
+
+  if ( $self->dba->get_division ne 'vertebrates' ) {
+    return( 1, "Display xrefs are not typically expected for non-vertebrates" );
+  }
+}
+
 sub tests {
   my ($self) = @_;
 

diff --git a/lib/Bio/EnsEMBL/DataCheck/Checks/HGNCMultipleGenes.pm b/lib/Bio/EnsEMBL/DataCheck/Checks/HGNCMultipleGenes.pm
@@ -31,7 +31,7 @@ extends 'Bio::EnsEMBL::DataCheck::DbCheck';
 use constant {
   NAME           => 'HGNCMultipleGenes',
   DESCRIPTION    => 'HGNC-derived gene names are not given to multiple genes',
-  GROUPS         => ['core', 'xref'],
+  GROUPS         => ['core', 'xref', 'xref_name_projection'],
   DATACHECK_TYPE => 'advisory',
   TABLES         => ['external_db', 'gene', 'xref']
 };

diff --git a/lib/Bio/EnsEMBL/DataCheck/index.json b/lib/Bio/EnsEMBL/DataCheck/index.json
@@ -599,7 +599,8 @@
       "description" : "Compare GO xref counts between two databases, categorised by source",
       "groups" : [
          "compare_core",
-         "xref"
+         "xref",
+         "xref_go_projection"
       ],
       "name" : "CompareGOXref",
       "package_name" : "Bio::EnsEMBL::DataCheck::Checks::CompareGOXref"
@@ -658,7 +659,8 @@
       "description" : "Compare GO xref counts between two databases, categorised by source coming from the info_type",
       "groups" : [
          "compare_core",
-         "xref"
+         "xref",
+         "xref_go_projection"
       ],
       "name" : "CompareProjectedGOXrefs",
       "package_name" : "Bio::EnsEMBL::DataCheck::Checks::CompareProjectedGOXrefs"
@@ -668,7 +670,8 @@
       "description" : "Compare Projected Gene Name counts between two databases",
       "groups" : [
          "compare_core",
-         "xref"
+         "xref",
+         "xref_name_projection"
       ],
       "name" : "CompareProjectedGeneNames",
       "package_name" : "Bio::EnsEMBL::DataCheck::Checks::CompareProjectedGeneNames"
@@ -678,7 +681,8 @@
       "description" : "Compare Projected Synonyms counts between two databases, categorised by db_name coming from the external_db",
       "groups" : [
          "compare_core",
-         "xref"
+         "xref",
+         "xref_name_projection"
       ],
       "name" : "CompareProjectedSynonyms",
       "package_name" : "Bio::EnsEMBL::DataCheck::Checks::CompareProjectedSynonyms"
@@ -1002,7 +1006,8 @@
       "description" : "At least one gene name exists",
       "groups" : [
          "core",
-         "xref"
+         "xref",
+         "xref_name_projection"
       ],
       "name" : "DisplayXrefExists",
       "package_name" : "Bio::EnsEMBL::DataCheck::Checks::DisplayXrefExists"
@@ -1339,7 +1344,8 @@
       "description" : "HGNC-derived gene names are not given to multiple genes",
       "groups" : [
          "core",
-         "xref"
+         "xref",
+         "xref_name_projection"
       ],
       "name" : "HGNCMultipleGenes",
       "package_name" : "Bio::EnsEMBL::DataCheck::Checks::HGNCMultipleGenes"