Feature/mvp

lib/Bio/EnsEMBL/DataCheck/Checks/BlankNulls.pm

@@ -57,13 +57,14 @@ sub tests {
 
   foreach my $nullable (@$nullables) {
     my ($table, $column) = @$nullable;
-
-    my $desc = "Nullable column $table.$column has no '' or 'NULL' string values";
-    my $sql  = qq/
-      SELECT COUNT(*) FROM $table
-      WHERE $column = '' OR $column = 'NULL'
-    /;
+    if ($table ne "meta"){
+	my $desc = "Nullable column $table.$column has no '' or 'NULL' string values";
+	my $sql  = qq/
+	    SELECT COUNT(*) FROM $table
+	    WHERE $column = '' OR $column = 'NULL'
+	    /;
     is_rows_zero($self->dba, $sql, $desc);
+    }
   }
 }
 

lib/Bio/EnsEMBL/DataCheck/Checks/ControlledMetaKeys.pm

@@ -36,22 +36,27 @@ use constant {
 
 sub tests {
   my ($self) = @_;
-
   my $species_id = $self->dba->species_id;
   my $group = $self->dba->group;
-
   my $sql = qq/
     SELECT meta_key, COUNT(*) FROM meta
     WHERE species_id = $species_id OR species_id IS NULL
     GROUP BY meta_key
   /;
+
   my $helper = $self->dba->dbc->sql_helper;
   my %meta_keys = %{ $helper->execute_into_hash(-SQL => $sql) };
 
+  #check target site is main / new and select mandatory metakeys 
+  my $filter_metakeys = '';
+  if (defined $self->target_site){
+    $filter_metakeys = " AND target_site like '\%".$self->target_site."\%' ";
+  }
+
   my $prod_sql = qq/
     SELECT name, is_optional
     FROM meta_key
-    WHERE FIND_IN_SET('$group', db_type) AND is_current = 1
+    WHERE FIND_IN_SET('$group', db_type) AND is_current = 1 $filter_metakeys
   /;
   my $prod_dba    = $self->get_dba('multi', 'production');
   my $prod_helper = $prod_dba->dbc->sql_helper;

lib/Bio/EnsEMBL/DataCheck/Checks/DisplayNameFormat.pm

@@ -40,7 +40,7 @@ sub tests {
   my $mca = $self->dba->get_adaptor("MetaContainer");
 
   # Check that the format of the display name conforms to expectations.
-  my $format = '[A-Za-z0-9\ ]+ \([A-Za-z0-9\(\)\/\-\_,\#\. ]+\) \- GCA_\d+\.\d+(?:\s\[[\w ]+\])?';
+  my $format = '[A-Za-z0-9\ ]+ \([A-Za-z0-9\(\)\/\-\_,\#\. ]+\) \- GC[AF]_\d+\.\d+(?:\s\[[\w ]+\])?';
 
   my $desc = "Display name has correct format";
   my $display_name = $mca->single_value_by_key('species.display_name');

lib/Bio/EnsEMBL/DataCheck/Checks/MetaKeyFormat.pm

@@ -44,7 +44,8 @@ sub tests {
   my %formats = (
     'annotation.provider_url'        => '(https?:\/\/.+|www.*\.ensembl\.org)',
     'assembly.provider_url'          => '(https?:\/\/.+|www.*\.ensembl\.org)',
-    'assembly.accession'             => 'GCA_\d+\.\d+',
+    'assembly.accession'             => 'GC[AF]_\d+\.\d+',
+    'assembly.alt_accession'         => 'GCA_\d+\.\d+',
     'assembly.date'                  => '\d{4}-\d{2}',
     'assembly.default'               => '[\w\.\-]+',
     'genebuild.id'                   => '\d+',

lib/Bio/EnsEMBL/DataCheck/Checks/SpeciesTaxonomy.pm

@@ -49,7 +49,7 @@ sub tests {
   # scientific name, to disambiguate in the case of multiple strains
   # or assemblies of the same species. Since the taxonomy database does
   # not always have that information, remove it before comparing.
-  $sci_name =~ s/ \(GCA_\d+\)//;
+  $sci_name =~ s/ \(GC[AF]_\d+\)//;
   $sci_name =~ s/ (str\.|strain) .*//;
 
   my $desc_1 = 'Species-related meta data exists';

lib/Bio/EnsEMBL/DataCheck/DbCheck.pm

@@ -57,6 +57,15 @@ subtype 'Registry', as 'Str', where {
 
 =head1 METHODS
 
+=head2 target_site
+  Description: Fetch mandatory meta keys based on target site 
+              current values are main/new
+=cut
+has 'target_site' => (
+  is  => 'ro',
+  isa => 'Str | Undef',
+);
+
 =head2 db_types
   Description: Database types for which this datacheck is appropriate.
 =cut

lib/Bio/EnsEMBL/DataCheck/Pipeline/DataCheckSubmission.pm

@@ -122,6 +122,7 @@ sub write_output {
     json_by_species => $self->param('json_by_species'),
 
     submission_job_id => $self->input_job->dbID,
+    target_site       => $self->param('target_site'),
   };
   $self->dataflow_output_id($params, 1);