configurable minimum base depth list for ampliconstats,

make amliconstats qc check to save coverage data

MANIFEST

@@ -1055,6 +1055,7 @@ t/data/portable_pipelines/ncov2019-artic-nf/cf01166c42a/product_release.yml
 t/data/portable_pipelines/ncov2019-artic-nf/cf01166c42a/product_release_no_pp.yml
 t/data/portable_pipelines/ncov2019-artic-nf/cf01166c42a/product_release_no_study.yml
 t/data/portable_pipelines/ncov2019-artic-nf/cf01166c42a/product_release_unknown_pp.yml
+t/data/portable_pipelines/ncov2019-artic-nf/cf01166c42a/product_release_explicit_astats_depth.yml
 t/data/portable_pipelines/ncov2019-artic-nf/v.3/product_release_no_staging_root.yml
 t/data/portable_pipelines/ncov2019-artic-nf/v.3/product_release_two_pps.yml
 t/data/portable_pipelines/ncov2019-artic-nf/v.3/product_release_two_studies.yml

lib/npg_pipeline/function/stage2pp.pm

@@ -33,8 +33,8 @@ Readonly::Hash   my %PER_PP_REQS   => (
   ncov2019_artic_nf => {memory_mb => 5000, num_cpus => 4},
   ncov2019_artic_nf_ampliconstats => {memory_mb => 1000, num_cpus => 2},
                                       );
-
-Readonly::Scalar my $AMPLICONSTATS_OPTIONS => q[-t 50 -d 1,10,20,100];
+Readonly::Array  my @DEFAULT_AMPLICONSTATS_DEPTH  => qw(1 10 20 100);
+Readonly::Scalar my $AMPLICONSTATS_OPTIONS        => q[-t 50];
 
 our $VERSION = '0';
 
@@ -311,6 +311,9 @@ sub _ncov2019_artic_nf_ampliconstats_create {
     return;
   }
 
+  my $depth_array = $pp->{'ampliconstats_min_base_depth'}
+                    || \@DEFAULT_AMPLICONSTATS_DEPTH;
+
   my $lane_product = ($product->lanes_as_products)[0];
   my $lane_pp_path = $self->pp_archive4product(
     $lane_product, $pp, $self->pp_archive_path());
@@ -344,20 +347,23 @@ sub _ncov2019_artic_nf_ampliconstats_create {
                                'ampliconstats',
                                $sta_cpus_option,
                                $AMPLICONSTATS_OPTIONS,
+                               q[-d ] . join(q[,], @{$depth_array}),
                                $self->_primer_bed_file($product),
                                $input_files_glob;
   $sta_command = join q[ > ], $sta_command, $sta_file;
   # Invoke a lane-level qc check on the ampliconstats file produced
   # in the previous step with an option to fan out qc check outputs
   # to individual per-sample directories.
+  my @sections = map {q[FPCOV-] . $_} @{$depth_array};
+  unshift @sections, q[FREADS];
   my $qca_command = join q[ ], $self->qc_cmd,
                                '--check generic',
                                '--spec ampliconstats',
                                '--rpt_list ' . $lane_product->composition->freeze2rpt,
                                '--input_files ' . $sta_file,
                                '--pp_name ncov2019_artic_nf_ampliconstats',
                                '--pp_version ' . $self->pp_version($pp),
-                               '--ampstats_section FREADS',
+                               (map {'--ampstats_section ' . $_} @sections),
                                '--qc_out ' . $lane_qc_dir,
                                '--sample_qc_out ' . q['] . $lane_archive . q[/plex*/qc'];
 

t/20-function-stage2pp.t

@@ -274,7 +274,7 @@ subtest 'skip unknown pipeline' => sub {
 };
 
 subtest q(definition generation, 'ncov2019_artic_nf ampliconstats' pp) => sub {
-  plan tests => 28;
+  plan tests => 30;
 
   local $ENV{NPG_CACHED_SAMPLESHEET_FILE} = q[t/data/samplesheet_33990.csv];
 
@@ -296,15 +296,25 @@ subtest q(definition generation, 'ncov2019_artic_nf ampliconstats' pp) => sub {
 
   my @commands = ();
   my @replacement_files = ();
-  for my $p ((1, 2)) {
+  my @astats_sections = qw(FREADS FPCOV-1 FPCOV-10 FPCOV-20 FPCOV-100);
+
+  my $count = 0;
+  for my $p ((1, 2, 1)) {
+
+    $count++;
+    my @s = @astats_sections;
+    $count == 3 and pop @s;
+    my $sections = join q[ ], map { q[--ampstats_section ] . $_ } @s;
+
     my $pp_path = qq(${pp_archive_path}/lane${p}) .
       qq(/ncov2019_artic_nf_ampliconstats/0.1/);
     my $astats_file = $pp_path . qq(26291_${p}.astats);
     my $replacement_map_file = $pp_path . q(replacement_map.txt);
     push @replacement_files, $replacement_map_file;
     push @commands,
                 '(' .
-      $dir . q(/samtools ampliconstats -@1 -t 50 -d 1,10,20,100 ) .
+      $dir . q(/samtools ampliconstats -@1 -t 50 -d 1,10,20) .
+      ($count == 3 ? q( ) : q(,100 )) .
       $dir . q(/primer_panel/nCoV-2019/default/SARS-CoV-2/MN908947.3/nCoV-2019.bed ) .
       $pp_archive_path . qq(/lane${p}) .
       q(/plex*/ncov2019_artic_nf/cf01166c42a) .
@@ -320,7 +330,7 @@ subtest q(definition generation, 'ncov2019_artic_nf ampliconstats' pp) => sub {
       $dir . q(/qc --check generic --spec ampliconstats ) .
       qq(--rpt_list 26291:${p} --input_files $astats_file ) .
       q(--pp_name ncov2019_artic_nf_ampliconstats --pp_version 0.1 ) .
-      q(--ampstats_section FREADS ) .
+      qq($sections ) .
       q(--qc_out ) . $archive_path . qq(/lane${p}/qc ) .
       q(--sample_qc_out ') . $archive_path . qq(/lane${p}/plex*/qc') .
                 ')';
@@ -379,6 +389,21 @@ subtest q(definition generation, 'ncov2019_artic_nf ampliconstats' pp) => sub {
   is ($d->job_name, 'stage2App_ncov20.1_26291', 'job name');
   is ($d->memory, 1000, 'memory');
   is_deeply ($d->num_cpus, [2], 'number of CPUs');
+
+  $ppd = npg_pipeline::function::stage2pp->new(
+    product_conf_file_path => qq[$repo_dir/product_release_explicit_astats_depth.yml],
+    pipeline_type          => 'stage2App',
+    archive_path           => $archive_path,
+    runfolder_path         => $runfolder_path,
+    id_run                 => 26291,
+    merge_lanes            => 0,
+    timestamp              => $timestamp,
+    repository             => $dir);
+
+  $ds = $ppd->create;
+  is (@{$ds}, 2, 'two definitions are returned');
+  $d = $ds->[0];
+  is_deeply ([(split q[ ], $d->command)], $commands[2], 'correct command for lane 1');
 };
 
 1;

t/data/portable_pipelines/ncov2019-artic-nf/cf01166c42a/product_release_explicit_astats_depth.yml

@@ -0,0 +1,26 @@
+---
+default:
+  s3:
+    enable: false
+    url: null
+    notify: false
+  irods:
+    enable: true
+    notify: false
+
+study:
+  - study_id: "3073"
+    portable_pipelines:
+      - pp_name: "ncov2019-artic-nf"
+        pp_version: "cf01166c42a"
+        pp_type: "stage2pp"
+        pp_root: "t/data/portable_pipelines"
+      - pp_name: "ncov2019-artic-nf ampliconstats"
+        pp_version: "0.1"
+        pp_type: "stage2App"
+        pp_input_glob: "plex*/ncov2019_artic_nf/cf01166c42a/ncovIlluminaCram_ncovIllumina_sequenceAnalysis_trimPrimerSequences/*primertrimmed.sorted.bam"
+        ampliconstats_min_base_depth:
+          - 1
+          - 10
+          - 20
+        pp_archival_flag: false