Merge pull request #877 from mgcam/daemon_pickup_illumina_only

Restricted pipeline daemons to deal with runs by a certain manufacturer.

Changes

@@ -1,6 +1,13 @@
 LIST OF CHANGES
 ---------------
 
+ - Changed pipeline daemons so that they pick up runs performed on the
+   instruments by a manufacturer with a name given by a new attribute -
+   instrument_name. By default this name is set to 'Illumina' so that
+   the daemons continue picking up Illumina-only runs even if other
+   runs are present in the run tracking database. 
+ - Deleted redundant t::dbic_util::test_schema_wh method.
+
 release 68.7.1 (2024-12-17)
  - Fix perlbrew installation by installing libdevel-patchperl-perl in runner
  - Changed/extended tests for npg_pipeline::function::autoqc to ensure that

MANIFEST

@@ -139,6 +139,7 @@ t/data/cache/my_samplesheet_12376.csv
 t/data/dbic_fixtures/000-Designation.yml
 t/data/dbic_fixtures/000-EntityType.yml
 t/data/dbic_fixtures/000-InstrumentStatusDict.yml
+t/data/dbic_fixtures/000-Manufacturer.yml
 t/data/dbic_fixtures/000-RunLaneStatusDict.yml
 t/data/dbic_fixtures/000-RunStatusDict.yml
 t/data/dbic_fixtures/000-Tag.yml

lib/npg_pipeline/daemon.pm

@@ -21,6 +21,7 @@ with qw{
 our $VERSION = '0';
 
 Readonly::Scalar my $SLEEPY_TIME => 900;
+Readonly::Scalar my $DEFAULT_INSTRUMENT_MANUFACTURER => 'Illumina';
 
 has 'pipeline_script_name' => (
   isa        => q{Str},
@@ -30,12 +31,21 @@ has 'pipeline_script_name' => (
   builder    => 'build_pipeline_script_name',
 );
 
+has 'manufacturer_name' => (
+  isa        => q{Str},
+  is         => q{ro},
+  default    => $DEFAULT_INSTRUMENT_MANUFACTURER,
+  documentation => 'Instrument manufacturer name, defaults to Illumina. ' .
+                   'Runs on instruments by this manufacturer only will be ' .
+                   'considered by this daemon.',
+);
+
 has 'dry_run' => (
   isa        => q{Bool},
   is         => q{ro},
   required   => 0,
   default    => 0,
-  documentation => 'dry run mode flag, false by default',
+  documentation => 'Dry run mode flag, false by default',
 );
 
 has 'seen' => (
@@ -58,6 +68,7 @@ sub _build_npg_tracking_schema {
 
 sub runs_with_status {
   my ($self, $status_name, $from_time) = @_;
+
   if (!$status_name) {
     $self->logcroak(q[Need status name]);
   }
@@ -73,7 +84,8 @@ sub runs_with_status {
   }
 
   return
-    map { $_->run() }
+    grep { $_->instrument_format->manufacturer->name eq $self->manufacturer_name }
+    map  { $_->run() }
     $self->npg_tracking_schema()->resultset(q[RunStatus])->search(
       $condition,
       {prefetch=>q[run_status_dict], order_by => q[me.date],}
@@ -194,8 +206,8 @@ case of error.
 =head2 runs_with_status
 
 With one argument, which should be a valid run status description,
-returns a list of DBIx::Class::Row objects from the Run result set,
-which correspond to runs with the current status descriptiongiven
+returns a list of C<DBIx::Class::Row> objects from the C<Run> result set,
+which correspond to runs with the current status description given
 by the argument.
   
   # find runs with current status 'archival pending'
@@ -215,6 +227,9 @@ should be after the time given by the second argument.
 In both cases a list of returned objects is sorted in the assending
 run status timestamp order.
 
+Only runs performed on the instruments by the manufacturer given by
+the C<manufacturer_name> attribute are chosen.
+
 If no run satisfies the conditions given by the argument(s), an
 empty list is returned.
 

t/50-npg_pipeline-daemon-analysis.t

@@ -1,6 +1,6 @@
 use strict;
 use warnings;
-use Test::More tests => 12;
+use Test::More tests => 8;
 use Test::Exception;
 use Cwd qw{ getcwd abs_path };
 use File::Path qw{ make_path };
@@ -30,19 +30,38 @@ my $schema = t::dbic_util->new()->test_schema();
 my $test_run = $schema->resultset(q[Run])->find(1234);
 $test_run->update_run_status('analysis pending', 'pipeline',);
 
-my %h = map { $_ => 1 } (1234, 4330, 4999, 5222);
-my @statuses = ();
-for my $r ($schema->resultset(q[Run])->search({})->all()) {
-  my $id = $r->id_run;
-  if ($h{$id}) {
-    cmp_ok ($r->current_run_status_description, 'eq',
-      'analysis pending', "test run $id is analysis pending");
-  } else {
-    push @statuses, $r->current_run_status_description;
+subtest 'runs and statuses' => sub {
+  plan tests => 10;
+
+  my %h = map { $_ => 1 } (1234, 4330, 4999, 5222, 100000);
+  my @other_statuses = ();
+  for my $r ($schema->resultset(q[Run])->search({})->all()) {
+    my $id = $r->id_run;
+    if ($h{$id}) {
+      cmp_ok ($r->current_run_status_description, 'eq',
+        'analysis pending', "test run $id is analysis pending");
+      } else {
+        push @other_statuses, $r->current_run_status_description;
+    }
   }
-}
-is (scalar(grep { $_ eq 'analysis pending' } @statuses), 0,
-  'other test runs are not analysis pending');
+  is (scalar(grep { $_ eq 'analysis pending' } @other_statuses), 0,
+    'other test runs are not analysis pending');
+
+  my $runner = npg_pipeline::daemon::analysis->new(npg_tracking_schema => $schema);
+  is ($runner->manufacturer_name, 'Illumina', 'default manufacturer name');
+  my @run_ids = sort { $a <=> $b } map { $_->id_run }
+    $runner->runs_with_status('analysis pending');
+  is (join(q[,], @run_ids), '1234,4330,4999,5222', 'correct Illumina runs IDs');
+
+  $runner = npg_pipeline::daemon::analysis->new(
+    manufacturer_name => 'Element Biosciences',
+    npg_tracking_schema => $schema
+  );
+  is ($runner->manufacturer_name, 'Element Biosciences',
+    'manufacturer name as set');
+  @run_ids = map { $_->id_run } $runner->runs_with_status('analysis pending');
+  ok ((@run_ids == 1) && ($run_ids[0] == 100000), 'One Element Biosciences run');
+};
 
 my $rf_path = '/some/path';
 

t/data/dbic_fixtures/000-Manufacturer.yml

@@ -0,0 +1,7 @@
+---
+- id_manufacturer: 20
+  name: Element Biosciences
+- id_manufacturer: 10
+  name: Illumina
+- id_manufacturer: 16
+  name: Roche/454

t/data/dbic_fixtures/100-InstrumentFormat.yml

@@ -30,3 +30,11 @@
   default_columns: 4
   days_between_washes: 28
   runs_between_washes: 1
+- id_instrument_format: 12
+  id_manufacturer: 20
+  model: AVITI23
+  iscurrent: 1
+  default_tiles: 0
+  default_columns: 0
+  days_between_washes: 30
+  runs_between_washes: 1

t/data/dbic_fixtures/200-Instrument.yml

@@ -106,4 +106,12 @@
   iscurrent: 1
   ipaddr: 
   instrument_comp: HS1
+- id_instrument: 70
+  name: AVITI2301
+  id_instrument_format: 12
+  external_name: ''
+  serial: ''
+  iscurrent: 1
+  ipaddr: ~ 
+  instrument_comp: ~
 

t/data/dbic_fixtures/300-Run.yml

@@ -379,4 +379,18 @@
   id_run_pair: ~
   is_paired: 0
   priority: 3
+  team: A
+- actual_cycle_count: 316
+  batch_id: ~
+  expected_cycle_count: 316
+  flowcell_id: ~
+  folder_name: 'elembio_run'
+  folder_path_glob: 't/data'
+  id_instrument: 70
+  id_instrument_format: 12
+  id_run: 100000
+  id_run_pair: ~
+  is_paired: 1
+  priority: 1
   team: A 
+

t/data/dbic_fixtures/400-RunStatus.yml

@@ -113,3 +113,22 @@
   id_run_status_dict: 4
   id_user: 7
   iscurrent: 1
+- id_run_status: 45
+  id_run: 100000
+  date: 2024-11-23 02:14:17
+  id_run_status_dict: 2
+  id_user: 7
+  iscurrent: 0
+- id_run_status: 46
+  id_run: 100000
+  date: 2024-11-23 12:23:10
+  id_run_status_dict: 4
+  id_user: 7
+  iscurrent: 0
+- id_run_status: 47
+  id_run: 100000
+  date: 2024-11-23 12:44:40
+  id_run_status_dict: 6
+  id_user: 7
+  iscurrent: 1
+

t/dbic_util.pm

@@ -25,13 +25,6 @@ sub test_schema_mlwh {
     );
 }
 
-sub test_schema_wh {
-    my ($self, $fixture_path) = @_;
-    return $self->create_test_db(
-        'npg_warehouse::Schema', $fixture_path
-    );
-}
-
 no Moose;
 __PACKAGE__->meta->make_immutable();
 1;