Merge pull request #843 from wtsi-npg/devel

pull from devel to master to create release 101.4.0

Changes

@@ -1,5 +1,11 @@
 LIST OF CHANGES
 
+release 101.4.0 (2024-08-30)
+ - Added species_from_reference_genome method to st::api::lims. If the
+   reference_genome value is defined, this new method returns the name of the
+   species. parse_reference_genome from npg_tracking::data::reference::find
+   is used for parsing the reference genome string.
+
 release 101.3.0 (2024-08-05)
  - Provided a direct access to the default section of the study configuration
    file.

lib/st/api/lims.pm

@@ -12,6 +12,7 @@ use Class::Load qw/load_class/;
 use npg_tracking::util::types;
 use npg_tracking::glossary::rpt;
 use npg_tracking::glossary::composition::factory::rpt_list;
+use npg_tracking::data::reference::find;
 
 our $VERSION = '0';
 
@@ -602,6 +603,34 @@ sub _build_reference_genome {
   return $rg;
 }
 
+=head2 species_from_reference_genome
+
+Extracts the species name from the value of the C<reference_genome> attribute
+and returns it. Returns an undefined value if the value of the C<reference_genome>
+attribute is not defined or if the the C<reference_genome> string does not match
+the expected pattern.
+
+Examples:
+
+ reference_genome: 'Homo_sapiens (GRCh38_full_analysis_set_plus_decoy_hla)'
+ species: 'Homo_sapiens'
+
+ reference_genome: 'Mus_musculus (GRCm38 + ensembl_84_transcriptome)'
+ species: 'Mus_musculus'
+=cut
+sub species_from_reference_genome {
+  my $self = shift;
+
+  if ($self->reference_genome) {
+    my @genome_as_array = npg_tracking::data::reference::find
+      ->parse_reference_genome($self->reference_genome);
+    if (@genome_as_array) {
+      return $genome_as_array[0];
+    }
+  }
+  return;
+}
+
 sub _trim_value {
   my ($self, $value) = @_;
   if ($value) {
@@ -1311,6 +1340,8 @@ __END__
 
 =item npg_tracking::glossary::composition::component::illumina
 
+=item npg_tracking::data::reference::find
+
 =back
 
 =head1 INCOMPATIBILITIES

t/40-st-lims.t

@@ -284,7 +284,7 @@ subtest 'Samplesheet driver for a one-component composition' => sub {
 };
 
 subtest 'Samplesheet driver for arbitrary compositions' => sub {
-  plan tests => 69;
+  plan tests => 75;
 
   my $path = 't/data/samplesheet/novaseq_multirun.csv';
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} = $path;
@@ -309,6 +309,8 @@ subtest 'Samplesheet driver for arbitrary compositions' => sub {
     is($o->library_name, '22802061', 'library name');
     is($o->reference_genome, 'Homo_sapiens (GRCh38_15_plus_hs38d1) [minimap2]',
       'reference genome');
+    is($o->species_from_reference_genome, 'Homo_sapiens',
+      'species reference genome');
   }
 
   $ss = $children[0];
@@ -422,7 +424,7 @@ subtest 'Insert size' => sub {
 };
 
 subtest 'Study and sample properties' => sub {
-  plan tests => 75;
+  plan tests => 83;
 
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} =
     't/data/samplesheet/4pool4libs_extended.csv';
@@ -442,6 +444,8 @@ subtest 'Study and sample properties' => sub {
   is( $lims->study_reference_genome, q[ ], 'study reference genome');
   is( $lims->reference_genome, 'Haemonchus_contortus (V1_21June13)',
     'reference genome');
+  is( $lims->species_from_reference_genome, 'Haemonchus_contortus',
+    'species from reference genome');
 
   # Individual plex.
   $lims = st::api::lims->new(id_run => 9999, position => 7, tag_index=> 76);
@@ -461,6 +465,8 @@ subtest 'Study and sample properties' => sub {
   is( $lims->study_reference_genome, 'Mus_musculus (GRCm38)',
     'study reference genome');
   is( $lims->reference_genome, 'Mus_musculus (GRCm38)', 'reference genome');
+  is( $lims->species_from_reference_genome, 'Mus_musculus',
+    'species from reference genome');
 
   # Indexed lane and tag zero for the same lane.
   for my $l (
@@ -475,6 +481,8 @@ subtest 'Study and sample properties' => sub {
     is( $l->study_reference_genome, 'Mus_musculus (GRCm38)',
       'study reference genome');
     is( $l->reference_genome, 'Mus_musculus (GRCm38)', 'reference genome');
+    is( $lims->species_from_reference_genome, 'Mus_musculus',
+      'species from reference genome');
   }
 
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} =
@@ -494,6 +502,8 @@ subtest 'Study and sample properties' => sub {
     is( $l->sample_reference_genome, undef, 'sample reference genome undefined');
     is( $l->study_reference_genome, $ref, 'study reference genome');
     is( $l->reference_genome, undef, 'no fallback to study');
+    is( $l->species_from_reference_genome, undef,
+      'species from reference genome is undefined');
   }
 
   my $ref2 = 'Homo_sapiens (GRCh38_full_analysis_set_plus_decoy_hla)';