PacBio iRODS data - set QC state metadata

... when it is safe to do so.

lib/WTSI/NPG/HTS/PacBio/Annotator.pm

@@ -225,6 +225,51 @@ sub make_tag_metadata {
   return $self->_make_multi_value_metadata(\@run_records, $method_attr);
 }
 
+=head2 make_qc_metadata
+
+  Arg [n]      PacBio run database records,
+               List[WTSI::DNAP::Warehouse::Schema::Result::PacBioRun].
++
+  Example    : my @avus = $ann->make_qc_metadata(@run_records);
+  Description: Return QC outcome AVU metadata for a single product.
+
+               An empty list is returned if the input list contains
+               either no records or multiple records or the only record
+               is not linked to a record in the pac_bio_product_metrics
+               table.
+
+               This method should be called in the context of a single
+               iRODS object. If, according to a record in the pac_bio_run
+               table, a well contains multiple samples, but in practice
+               no deplexing was done, when trying to establish data
+               provenance we might get multiple pac_bio_run table rows.
+               Opting out of assigning a QC outcome in this case is
+               a conscious conservative decision that was made at the
+               time of writing (March 2024).
+
+  Returntype : List[HashRef]
+
+=cut
+
+sub make_qc_metadata {
+  my ($self, @run_records) = @_;
+
+   my @avus = ();
+  if (@run_records == 1) {
+    my @product_metrics = $run_records[0]->pac_bio_product_metrics()->all();
+    # Absence of linked product records is not unknown, one linked product
+    # record is normal, multiple linked records is, most likely, an error.
+    if (@product_metrics == 1) {
+      my $qc_outcome = $product_metrics[0]->qc();
+      if (defined $qc_outcome) {
+        push @avus, $self->make_avu($QC_STATE, $qc_outcome);
+      }
+    }
+  }
+
+  return @avus;
+}
+
 sub _make_multi_value_metadata {
   my ($self, $objs, $method_attr) = @_;
   # The method_attr argument is a map of method name to attribute name

lib/WTSI/NPG/HTS/PacBio/MetaUpdater.pm

@@ -78,7 +78,8 @@ sub update_secondary_metadata {
     try {
       my @run_records = $self->find_pacbio_runs(
           $id_run, $well, $tag_id, $plate_number);
-      my @secondary_avus = $self->make_secondary_metadata(@run_records);
+      my @secondary_avus = map { $self->$_(@run_records) }
+        qw/make_secondary_metadata make_qc_metadata/;
       $obj->update_secondary_metadata(@secondary_avus);
 
       $self->info("Updated metadata on '$path' ",

t/lib/WTSI/NPG/HTS/PacBio/MetaUpdaterTest.pm

@@ -73,7 +73,7 @@ sub require : Test(1) {
   require_ok('WTSI::NPG::HTS::PacBio::MetaUpdater');
 }
 
-sub update_secondary_metadata : Test(7) {
+sub update_secondary_metadata : Test(8) {
   my $irods = WTSI::NPG::iRODS->new(environment          => \%ENV,
                                     strict_baton_version => 0);
   my $updater = WTSI::NPG::HTS::PacBio::MetaUpdater->new
@@ -135,7 +135,20 @@ sub update_secondary_metadata : Test(7) {
     'plate_number', $plate_number); # Update primary metadata
   my $new_study_name = 'Updated for the test';
   $study_row->update({name => $new_study_name}); # Update study name in mlwh
-
+  # Create a linked product row with QC outcome defined.
+  my $rw_row = $wh_schema->resultset('PacBioRunWellMetric')->create({
+    pac_bio_run_name => $run_name,
+    well_label => $well_label,
+    plate_number => $plate_number,
+    id_pac_bio_product => 'A' x 64,
+    instrument_type => 'SomeType'
+  });
+  my $p_row = $wh_schema->resultset('PacBioProductMetric')->create({
+    id_pac_bio_rw_metrics_tmp => $rw_row->id_pac_bio_rw_metrics_tmp,
+    id_pac_bio_tmp => $row->id_pac_bio_tmp,
+    id_pac_bio_product => $rw_row->id_pac_bio_product
+  });
+
   # Call the updater again.
   $updater->update_secondary_metadata(\@paths_to_update);
 
@@ -165,6 +178,25 @@ sub update_secondary_metadata : Test(7) {
 
   is_deeply($updated_as_dict, $expected_as_dict,
     'Updated metadata is correct after the update');
+
+  $p_row->update({qc => 1});
+  $expected_as_dict->{$QC_STATE} = 1;
+  for my $key ((map { $_ . '_history'} ($STUDY_NAME, $PACBIO_STUDY_NAME))) {
+    delete $expected_as_dict->{$key};
+  }
+  # Call the updater again.
+  $updater->update_secondary_metadata(\@paths_to_update); 
+  # Inspect the metadata after the update.
+  $updated = WTSI::NPG::HTS::DataObject->new
+    (collection  => $irods_tmp_coll,
+     data_object => $data_file,
+     irods       => $irods)->metadata;
+
+  $updated_as_dict = {};
+  for my $meta (@{$updated}) {
+    $updated_as_dict->{$meta->{'attribute'}} = $meta->{'value'};
+  }
+  is($expected_as_dict->{$QC_STATE}, 1, 'Updated metadata contains qc outcome');  
 }
 
 1;