uncouple deepvariant from gatk combinevariants (#1843)

* uncouple deepvariant from combinevariants * Add CLI option to combine vcf output from specific recipes * Update variable names Glnexus recipe * Fix github action errors * Fix github action errors * Fix github action error
Clinical-Genomics · Mar 23, 2021 · 4fedaaf · 4fedaaf
1 parent 8c31b90
commit 4fedaaf
Show file tree

Hide file tree

Showing 11 changed files with 73 additions and 23 deletions.
diff --git a/definitions/rd_dna_panel_parameters.yaml b/definitions/rd_dna_panel_parameters.yaml
@@ -759,6 +759,12 @@ gatk_combinevariants_prioritize_caller:
   data_type: SCALAR
   mandatory: no
   type: recipe_argument
+gatk_combinevariants_callers_to_combine:
+  associated_recipe:
+    - gatk_combinevariantcallsets
+  data_type: ARRAY
+  mandatory: no
+  type: recipe_argument
 ## bcftools norm
 bcftools_norm:
   analysis_mode: case

diff --git a/definitions/rd_dna_parameters.yaml b/definitions/rd_dna_parameters.yaml
@@ -534,7 +534,7 @@ gatk_baserecalibration:
   associated_recipe:
     - mip
   data_type: SCALAR
-  default: 0
+  default: 1
   file_tag: _brecal
   outfile_suffix: ".bam"
   program_executables:
@@ -1201,7 +1201,7 @@ gatk_haplotypecaller:
   associated_recipe:
     - mip
   data_type: SCALAR
-  default: 0
+  default: 1
   file_tag: _haptc
   outfile_suffix: ".vcf"
   program_executables:
@@ -1272,7 +1272,7 @@ gatk_genotypegvcfs:
   associated_recipe:
     - mip
   data_type: SCALAR
-  default: 0
+  default: 1
   file_tag: _gent
   outfile_suffix: ".vcf"
   program_executables:
@@ -1297,7 +1297,7 @@ gatk_gathervcfs:
   associated_recipe:
     - mip
   data_type: SCALAR
-  default: 0
+  default: 1
   file_tag: ""
   outfile_suffix: ".vcf"
   program_executables:
@@ -1315,7 +1315,7 @@ gatk_variantrecalibration:
   associated_recipe:
     - mip
   data_type: SCALAR
-  default: 0
+  default: 1
   file_tag: _vrecal
   program_executables:
     - bcftools
@@ -1470,6 +1470,12 @@ gatk_combinevariants_prioritize_caller:
   data_type: SCALAR
   mandatory: no
   type: recipe_argument
+gatk_combinevariants_callers_to_combine:
+  associated_recipe:
+    - gatk_combinevariantcallsets
+  data_type: ARRAY
+  mandatory: no
+  type: recipe_argument
 prepareforvariantannotationblock:
   analysis_mode: case
   associated_recipe:

diff --git a/lib/MIP/Cli/Mip/Analyse/Rd_dna.pm b/lib/MIP/Cli/Mip/Analyse/Rd_dna.pm
@@ -1403,6 +1403,15 @@ q{Number of hom-ref genotypes to infer at sites not present in a panel. Connecte
         )
     );
 
+    option(
+        q{gatk_combinevariants_callers_to_combine} => (
+            cmd_flag      => q{gatk_combinevar_use_callers},
+            documentation => q{Combine vcf output from these recipes},
+            is            => q{rw},
+            isa           => enum( [qw{ gatk_variantcalibration glnexus_merge }] ),
+        )
+    );
+
     option(
         q{gatk_variantevalall} => (
             cmd_tags      => [q{Analysis recipe switch}],

diff --git a/lib/MIP/Cli/Mip/Analyse/Rd_dna_panel.pm b/lib/MIP/Cli/Mip/Analyse/Rd_dna_panel.pm
@@ -835,6 +835,15 @@ q{Number of hom-ref genotypes to infer at sites not present in a panel. Connecte
         )
     );
 
+    option(
+        q{gatk_combinevariants_callers_to_combine} => (
+            cmd_flag      => q{gatk_combinevar_use_callers},
+            documentation => q{Combine vcf output from these recipes},
+            is            => q{rw},
+            isa           => enum( [qw{ gatk_variantcalibration glnexus_merge }] ),
+        )
+    );
+
     option(
         q{gatk_variantevalall} => (
             cmd_tags      => [q{Analysis recipe switch}],

diff --git a/lib/MIP/Recipes/Analysis/Gatk_combinevariantcallsets.pm b/lib/MIP/Recipes/Analysis/Gatk_combinevariantcallsets.pm
@@ -135,13 +135,13 @@ sub analysis_gatk_combinevariantcallsets {
     my @variant_callers;
 
     ## Only process active callers
-  CALLER:
-    foreach my $variant_caller ( @{ $parameter_href->{cache}{variant_callers} } ) {
-        if ( $active_parameter_href->{$variant_caller} ) {
-
-            push @variant_callers, $variant_caller;
+    CALLER:
+      foreach my $variant_caller ( @{ $active_parameter_href->{gatk_combinevariants_callers_to_combine} } ) {
+          if ( $active_parameter_href->{$variant_caller} ) {
+
+                push @variant_callers, $variant_caller;
+          }
         }
-    }
 
     ## Stores the parallel chains that job ids should be inherited from
     my @parallel_chains;

diff --git a/lib/MIP/Recipes/Analysis/Glnexus.pm b/lib/MIP/Recipes/Analysis/Glnexus.pm
@@ -118,7 +118,7 @@ sub analysis_glnexus {
     use MIP::Program::Htslib qw{ htslib_bgzip };
     use MIP::Processmanagement::Processes qw{ submit_recipe };
     use MIP::Recipe qw{ parse_recipe_prerequisites };
-    use MIP::Sample_info qw{ set_recipe_outfile_in_sample_info };
+    use MIP::Sample_info qw{ set_file_path_to_store set_recipe_outfile_in_sample_info };
     use MIP::Script::Setup_script qw{ setup_script };
 
     ### PREPROCESSING:
@@ -173,7 +173,7 @@ sub analysis_glnexus {
     );
 
     my $outfile_path_prefix = $io{out}{file_path_prefix};
-    my $outfile_path =
+    my $temp_outfile_path =
       catdir( $active_parameter_href->{temp_directory}, $io{out}{file_name_prefix} . q{.vcf} );
 
     ## Filehandles
@@ -211,7 +211,7 @@ sub analysis_glnexus {
                 filehandle       => $filehandle,
                 infile_paths_ref => \@genotype_infile_paths,
                 memory           => $memory,
-                stdoutfile_path  => $outfile_path,
+                stdoutfile_path  => $temp_outfile_path,
                 threads          => $core_number,
             }
         );
@@ -223,7 +223,7 @@ sub analysis_glnexus {
             {
                 filehandle          => $filehandle,
                 index_type          => q{tbi},
-                infile_path         => $outfile_path,
+                infile_path         => $temp_outfile_path,
                 outfile_path_prefix => $outfile_path_prefix,
                 output_type         => q{z},
                 threads             => $core_number,
@@ -251,7 +251,18 @@ sub analysis_glnexus {
 
         set_recipe_outfile_in_sample_info(
             {
-                path             => $outfile_path,
+                path             => $outfile_path_prefix . q{.vcf.gz},
+                recipe_name      => $recipe_name,
+                sample_info_href => $sample_info_href,
+            }
+        );
+
+        set_file_path_to_store(
+            {
+                format           => q{vcf},
+                id               => $case_id,
+                path             => $outfile_path_prefix . q{.vcf.gz},
+                path_index       => $outfile_path_prefix . q{.vcf.gz.tbi},
                 recipe_name      => $recipe_name,
                 sample_info_href => $sample_info_href,
             }

diff --git a/lib/MIP/Recipes/Pipeline/Analyse_rd_dna_panel.pm b/lib/MIP/Recipes/Pipeline/Analyse_rd_dna_panel.pm
@@ -232,8 +232,8 @@ sub parse_rd_dna_panel {
     ## Check that all active variant callers have a prioritization order and that the prioritization elements match a supported variant caller
     parse_prioritize_variant_callers(
         {
-            active_parameter_href => $active_parameter_href,
-            parameter_href        => $parameter_href,
+           active_parameter_href => $active_parameter_href,
+           parameter_href        => $parameter_href,
         }
     );
 

diff --git a/t/analysis_gatk_combinevariantcallsets.t b/t/analysis_gatk_combinevariantcallsets.t
@@ -65,11 +65,11 @@ $active_parameter{gatk_variantrecalibration}        = 1;
 $active_parameter{glnexus_merge}                    = 1;
 $active_parameter{recipe_core_number}{$recipe_name} = 1;
 $active_parameter{recipe_time}{$recipe_name}        = 1;
+$active_parameter{gatk_combinevariants_callers_to_combine} = [ qw( gatk_variantrecalibration glnexus_merge ) ];
 my $case_id = $active_parameter{case_id};
 $active_parameter{gatk_path}                            = q{gatk.jar};
 $active_parameter{gatk_combinevariantcallsets_bcf_file} = 1;
 
-my @variant_callers = qw{ gatk_variantrecalibration glnexus_merge };
 
 my %file_info = test_mip_hashes(
     {
@@ -86,7 +86,7 @@ my %parameter = test_mip_hashes(
 );
 
 CALLER:
-foreach my $caller (@variant_callers) {
+foreach my $caller ( @{ $active_parameter{gatk_combinevariants_callers_to_combine} } ) {
 
     test_add_io_for_recipe(
         {
@@ -133,7 +133,7 @@ my $is_ok = analysis_gatk_combinevariantcallsets(
 ok( $is_ok, q{ Executed analysis recipe } . $recipe_name . q{ multiple callers} );
 
 ## Given analysis parameters and single callers
-@{ $parameter{cache}{variant_callers} } = $parameter{cache}{variant_callers}[0];
+$active_parameter{gatk_combinevariants_callers_to_combine} = [ qw{ gatk_variantrecalibration } ];
 $is_ok = analysis_gatk_combinevariantcallsets(
     {
         active_parameter_href => \%active_parameter,

diff --git a/templates/grch38_mip_rd_dna_config.yaml b/templates/grch38_mip_rd_dna_config.yaml
@@ -66,7 +66,10 @@ vcfanno_config: grch38_vcfanno_config_-v0.1-.toml
 ### Analysis
 ## Programs
 ## Parameters
-gatk_combinevariants_prioritize_caller: deepvariant
+gatk_combinevariants_prioritize_caller: deepvariant,haplotypecaller
+gatk_combinevariants_callers_to_combine:
+  - gatk_variantrecalibration
+  - glnexus_merge
 gatk_path: /opt/conda/opt/gatk-3.8
 qccollect_sampleinfo_file: cluster_constant_path!/case_id!/analysis_constant_path!/case_id!_qc_sample_info.yaml
 picardtools_path: /usr/picard/

diff --git a/templates/mip_rd_dna_config.yaml b/templates/mip_rd_dna_config.yaml
@@ -47,7 +47,10 @@ fqf_annotations:
   - GNOMADAF
   - GNOMADAF_popmax
   - SWEGENAF
-gatk_combinevariants_prioritize_caller: deepvariant
+gatk_combinevariants_prioritize_caller: deepvariant,haplotypecaller
+gatk_combinevariants_callers_to_combine:
+  - gatk_variantrecalibration
+  - glnexus_merge
 gatk_path: /usr
 picardtools_path: /usr/picard
 qccollect_sampleinfo_file: cluster_constant_path!/case_id!/analysis_constant_path!/case_id!_qc_sample_info.yaml

diff --git a/templates/mip_rd_dna_panel_config.yaml b/templates/mip_rd_dna_panel_config.yaml
@@ -33,6 +33,9 @@ fqf_annotations:
   - GNOMADAF_popmax
   - SWEGENAF
 gatk_combinevariants_prioritize_caller: haplotypecaller
+gatk_combinevariants_callers_to_combine:
+  - gatk_variantrecalibration
+  - glnexus_merge
 picardtools_path: /usr/picard
 qccollect_sampleinfo_file: cluster_constant_path!/case_id!/analysis_constant_path!/case_id!_qc_sample_info.yaml
 vep_custom_annotation: