Merge pull request #861 from marcomoscasgr/feature/sample-metadata-in…

…-lims

Feature: sample metadata in lims

MANIFEST

@@ -667,6 +667,7 @@ t/data/samplesheet/samplesheet_33990.csv
 t/data/samplesheet/samplesheet_47995.csv
 t/data/samplesheet/samplesheet_singlecell_48460.csv
 t/data/samplesheet/4pool4libs_extended.csv
+t/data/samplesheet/4pool4libs_extended2.csv
 t/data/samplesheet/6946_extended.csv
 t/data/samplesheet/7007_extended.csv
 t/data/samplesheet/8pools_extended.csv

lib/st/api/lims.pm

@@ -285,6 +285,7 @@ Readonly::Hash my %ATTRIBUTE_LIST_METHODS => {
                            public_name
                            reference_genome
                            supplier_name
+                           uuid
                          /],
     'study'        => [qw/ accession_number
                            id
@@ -333,6 +334,10 @@ Similar to sample_names, but for public_names.
 
 Similar to sample_names, but for supplier_names.
 
+=head2 sample_uuids
+
+Similar to sample_names, but for LIMS-specific UUIDs
+
 =head2 study_names
 
 A list of study names. if $self->is_pool is true, returns unique study

t/40-st-lims-merge.t

@@ -75,7 +75,7 @@ subtest 'Create lane object from plex object' => sub {
 };
 
 subtest 'Create tag zero object' => sub {
-  plan tests => 10;
+  plan tests => 14;
 
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} =
     't/data/test40_lims/samplesheet_novaseq4lanes.csv';
@@ -124,8 +124,11 @@ subtest 'Create tag zero object' => sub {
     is ($t0->driver->mlwh_schema, $schema_wh,
       'the original db connection is retained');
     my @names = $t0->sample_names();
+    my @uuids = $t0->sample_uuids();
     is (@names, 18, '18 sample names are retrieved');
+    is (@uuids, 18, '18 sample uuids are retrieved');
     is ($names[0], '6751STDY13219539', 'first sample name is correct');
+    is ($uuids[0], '5832d018-56a6-11ed-a8fb-fa163eea3084', 'first sample uuid is correct');
   }
 };
 

t/40-st-lims.t

@@ -320,7 +320,7 @@ subtest 'Samplesheet driver for a one-component composition' => sub {
 };
 
 subtest 'Samplesheet driver for arbitrary compositions' => sub {
-  plan tests => 95;
+  plan tests => 108;
 
   my $path = 't/data/samplesheet/novaseq_multirun.csv';
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} = $path;
@@ -333,6 +333,7 @@ subtest 'Samplesheet driver for arbitrary compositions' => sub {
   is ($ss->tag_index, undef, 'tag_index undefined');
   ok (!$ss->is_pool, 'not a pool');
   is ($ss->is_composition, 1, 'this is a composition');
+  is ($ss->sample_uuid, undef, 'sample uuid');
   my @children = $ss->children();
   is (scalar @children, 5, 'five children');
 
@@ -341,8 +342,7 @@ subtest 'Samplesheet driver for arbitrary compositions' => sub {
     is($o->default_tagtwo_sequence, 'CCAACAGA', 'tag2 sequence');
     is($o->default_library_type, 'HiSeqX PCR free', 'library type');
     is($o->sample_name, '7592352', 'sample name');
-    is($o->sample_uuid, undef, 'sample uuid');
-    is($o->sample_lims, undef, 'sample lims');
+    is($o->sample_lims, 'SQSCP', 'sample lims');
     is($o->study_name, 'UK Study', 'study name');
     is($o->library_name, '22802061', 'library name');
     is($o->reference_genome, 'Homo_sapiens (GRCh38_15_plus_hs38d1) [minimap2]',
@@ -360,26 +360,54 @@ subtest 'Samplesheet driver for arbitrary compositions' => sub {
   is ($ss->tag_index, 9, 'correct tag_index');
   ok (!$ss->is_pool, 'plex is not a pool');
   is ($ss->is_composition, 0, 'not a composition');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b890', 'child sample uuid');
 
   $ss = $children[1];
   is ($ss->id_run, 26480, 'correct run id');
   is ($ss->position, 2, 'correct position');
   is ($ss->tag_index, 9, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b890', 'child sample uuid');
 
   $ss = $children[2];
   is ($ss->id_run, 26480, 'correct run id');
   is ($ss->position, 3, 'correct position');
   is ($ss->tag_index, 9, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b890', 'child sample uuid');
 
   $ss = $children[3];
   is ($ss->id_run, 26480, 'correct run id');
   is ($ss->position, 4, 'correct position');
   is ($ss->tag_index, 9, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b890', 'child sample uuid');
 
   $ss = $children[4];
   is ($ss->id_run, 28780, 'correct run id');
   is ($ss->position, 2, 'correct position');
   is ($ss->tag_index, 4, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b892', 'child sample uuid');
+
+  $rpt_list = '28780:1:4;28780:3:4;28780:4:4';
+  $ss=st::api::lims->new(rpt_list => $rpt_list);
+  is ($ss->sample_uuid, undef, 'sample uuid');
+
+  @children = $ss->children();
+  $ss = $children[0];
+  is ($ss->id_run, 28780, 'correct run id');
+  is ($ss->position, 1, 'correct position');
+  is ($ss->tag_index, 4, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b891', 'child sample uuid');
+
+  $ss = $children[1];
+  is ($ss->id_run, 28780, 'correct run id');
+  is ($ss->position, 3, 'correct position');
+  is ($ss->tag_index, 4, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b893', 'child sample uuid');
+
+  $ss = $children[2];
+  is ($ss->id_run, 28780, 'correct run id');
+  is ($ss->position, 4, 'correct position');
+  is ($ss->tag_index, 4, 'correct tag_index');
+  is ($ss->sample_uuid, 'a12b3456-a1bb-12a3-1234-12a34567b894', 'child sample uuid');
 
   $path = 't/data/samplesheet/data4merge.csv';
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} = $path;
@@ -491,10 +519,10 @@ subtest 'Insert size' => sub {
 };
 
 subtest 'Study and sample properties' => sub {
-  plan tests => 83;
+  plan tests => 97;
 
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} =
-    't/data/samplesheet/4pool4libs_extended.csv';
+    't/data/samplesheet/4pool4libs_extended2.csv';
 
   # A simple non-indexed lane.
   my $lims = st::api::lims->new(id_run => 9999, position => 1);
@@ -513,6 +541,9 @@ subtest 'Study and sample properties' => sub {
     'reference genome');
   is( $lims->species_from_reference_genome, 'Haemonchus_contortus',
     'species from reference genome');
+  my $sample_uuid = $lims->sample_uuid;
+  is_deeply([$lims->sample_uuids], [$sample_uuid],
+    'sample uuids for non-indexed lane');
 
   # Individual plex.
   $lims = st::api::lims->new(id_run => 9999, position => 7, tag_index=> 76);
@@ -534,8 +565,15 @@ subtest 'Study and sample properties' => sub {
   is( $lims->reference_genome, 'Mus_musculus (GRCm38)', 'reference genome');
   is( $lims->species_from_reference_genome, 'Mus_musculus',
     'species from reference genome');
+  $sample_uuid = $lims->sample_uuid;
+  is_deeply([$lims->sample_uuids], [$sample_uuid], 'sample uuids for a plex');
+  is( $lims->sample_lims, 'SQSCP', 'sample lims');
 
-  # Indexed lane and tag zero for the same lane.
+  my $sample_uuid_77 = st::api::lims->new(
+    id_run => 9999, position => 7, tag_index=> 77)->sample_uuid;
+
+  # Indexed lane and tag zero for the same lane, sample_lims and
+  # sample_uuid are undefined for tag 168 (control), one-sample pool.
   for my $l (
     st::api::lims->new(id_run => 9999, position => 7),
     st::api::lims->new(id_run => 9999, position => 7, tag_index => 0)
@@ -548,10 +586,28 @@ 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',
+    is( $l->species_from_reference_genome, 'Mus_musculus',
       'species from reference genome');
+    is_deeply( [$l->sample_uuids(1)], [$sample_uuid, $sample_uuid_77],
+      'sample uuids for a lane or tag0, counting controls');
+    is_deeply( [$l->sample_uuids(0)], [$sample_uuid, $sample_uuid_77],
+      'sample uuids for a lane or tag0, not counting the control');
+    is( $l->sample_lims, 'SQSCP', 'sample lims');
   }
 
+  # Tag zero, multiple plexes.
+  $lims = st::api::lims->new(id_run => 6946, position => 4, tag_index => 0);
+  is( $lims->sample_uuid, undef, 'sample uuid is undefined for tag0');
+  is( $lims->sample_lims, undef,
+    'sample lims is undefined for tag0 where samples come from different LIMS');
+  my $num_plexes = scalar $lims->children();
+  is( scalar $lims->sample_uuids(1), $num_plexes,
+    'multiple uuids, counting controls');
+  is( scalar $lims->sample_uuids(), $num_plexes,
+    'multiple uuids, defalt invocation - counting controls'); 
+  is( scalar $lims->sample_uuids(0), $num_plexes-1,
+    'multiple uuids, not counting controls'); 
+
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} =
     't/data/samplesheet/samplesheet_47539.csv';