Merge pull request #591 from wtsi-npg/devel

mege devel into master to create release 66.1

Changes

@@ -1,5 +1,10 @@
 LIST OF CHANGES FOR NPG-QC PACKAGE
 
+release 66.1
+ - autoqc results loader to check for potential composition digest clashes
+ - amend upstream_tags check to find previous runs which have reached
+     status "analysis complete" (instead of "run complete")
+
 release 66.0
  - GUI for utility QC (inactive)
  - add new bcfstats qc check + viewer change

lib/npg_qc/Schema/ResultSet.pm

@@ -123,7 +123,7 @@ sub find_seq_composition {
 }
 
 sub find_or_create_seq_composition {
-  my ($self, $composition) = @_;
+  my ($self, $composition, $check_digest) = @_;
 
   $self->_validate($composition);
   my $digest         = $composition->digest;
@@ -133,7 +133,14 @@ sub find_or_create_seq_composition {
     my $composition_row = $self->_find_composition($digest, $num_components);
     # If composition exists, we assume that it's properly defined, i.e.
     # all relevant components and records in the linking table exist.
-    if (!$composition_row) {
+    if ($composition_row) {
+      # If asked, ensure that the found row represents the same composition.
+      if ($check_digest &&
+           ($composition->freeze ne $composition_row->create_composition()->freeze)) {
+        croak "A different composition object with the same digest '$digest' " .
+              'already exists in the database';
+      }
+    } else {
       $composition_row = $schema->resultset('SeqComposition')
                                 ->create({
                     'digest' => $digest,
@@ -306,6 +313,10 @@ for every component-composition pair.
 Gives an error if this resultset does not have a relationship to
 the seq_composition table.
 
+If a true value is given as a second attrubute and a composition object is
+found, a check for a clash of sha256 digest of composition objects is performed.
+Error is raised if the clashing digests are detected.
+
 =head2 composition_fk_column_name
 
  Returns the name of the column with a foreign key linking this table to the

lib/npg_qc/autoqc/checks/upstream_tags.pm

@@ -608,7 +608,7 @@ sub _fetch_run_rows {
     where
       r.id_instrument = i.id_instrument
       and r.id_run = rs.id_run
-      and rs.id_run_status_dict = 4
+      and rs.id_run_status_dict = 9
       and i.id_instrument = (select id_instrument from run where id_run = ?)
       and rs.date <= (select max(date) from run_status where id_run = ? and id_run_status_dict in (2,3,4))
       and if(tr.id_tag=22, 'A', IF(tr.id_tag=23, 'B', 'X')) = (select if(id_tag=22, 'A', IF(id_tag=23, 'B', 'X')) from (run rx left outer join tag_run trx on rx.id_run = trx.id_run and trx.id_tag in (22,23)) where rx.id_run = ? and (trx.id_tag in (22,23) or trx.id_tag is null))

lib/npg_qc/autoqc/db_loader.pm

@@ -244,7 +244,8 @@ sub _json2db{
     if ($dbix_class_name && $self->_schema_has_source($dbix_class_name) &&
         $self->_pass_filter($obj)) {
       my $rs  = $self->schema->resultset($dbix_class_name);
-      my $related_composition = $rs->find_or_create_seq_composition($obj->composition());
+      my $check_digest = 1;
+      my $related_composition = $rs->find_or_create_seq_composition($obj->composition(), $check_digest);
       if (!$related_composition) {
         croak 'Composition is not found/created';
       }

t/50-schema-resultset.t

@@ -530,7 +530,7 @@ subtest q[not saving composition subset value "all"] => sub {
 my $id_run;
 
 subtest q[finding existing composition] => sub {
-  plan tests => 15;
+  plan tests => 17;
 
   my $rs_component   = $schema->resultset('SeqComponent');
   my $rs_composition = $schema->resultset('SeqComposition');
@@ -592,12 +592,25 @@ subtest q[finding existing composition] => sub {
         $f->add_component(npg_tracking::glossary::composition::component::illumina
                           ->new(%{$value}, tag_index => undef, subset => undef));
       }
-      $row = $ssrs->find_or_create_seq_composition($f->create_composition());
+      my $c = $f->create_composition();
+      $row = $ssrs->find_or_create_seq_composition($c);
       is($row->id_seq_composition, $pk_value, 'existing row returned');
+
+      my $row_to_change = $rs_component->search(
+        {id_run => $id_run,position => 1,tag_index => undef, subset => undef})->next();
+      $row_to_change or die 'Failed to find existing row';
+      $row_to_change->update({subset => 'human'});
+      my $d = $c->digest;
+      throws_ok {$ssrs->find_or_create_seq_composition($c, 1)}
+        qr/A different composition object with the same digest '$d' already exists in the database/,
+        'error detecting composition digest clash';
+      $row_to_change->update({subset => undef});
+      lives_ok {$ssrs->find_or_create_seq_composition($c, 1)}
+        'no error after removing the clash';
     }
 
     $count++;
-  }
+  }  
 };
 
 subtest q[creating new composition from new and existing components] => sub {