GATK CNVCaller fixes

CHANGELOG.md

@@ -19,6 +19,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - ngsbits samplegender to check sex [#453](https://github.com/nf-core/raredisease/pull/453)
 - New workflow for generating cgh files from SV vcfs for interpretation in the CytosSure interpretation software. Turned off by default [#456](https://github.com/nf-core/raredisease/pull/456/)
 - Fastp to do adapter trimming. It can be skipped using `--skip_fastp` [#457](https://github.com/nf-core/raredisease/pull/457)
+- GATK CNVCaller uses segments instead of intervals, filters out "reference" segments between the calls, and fixes a bug with how `ch_readcount_intervals` was handled
 
 ### `Changed`
 

conf/modules/call_sv_germlinecnvcaller.config

@@ -37,5 +37,10 @@ process {
             ext.args = "--run-mode CASE"
             ext.prefix = { "${meta.id}_${model.simpleName}" }
         }
+
+        withName: '.*CALL_STRUCTURAL_VARIANTS:CALL_SV_GERMLINECNVCALLER:BCFTOOLS_VIEW' {
+            ext.prefix = { "${meta.id}_gatkcnv_segments_reffiltered" }
+            ext.args = { '--output-type z --exclude "N_ALT = 0" ' }
+        }
     }
 }

subworkflows/local/call_structural_variants.nf

@@ -46,7 +46,7 @@ workflow CALL_STRUCTURAL_VARIANTS {
 
         if (!params.skip_germlinecnvcaller) {
             CALL_SV_GERMLINECNVCALLER (ch_genome_bam_bai, ch_genome_fasta, ch_genome_fai, ch_readcount_intervals, ch_genome_dictionary, ch_ploidy_model, ch_gcnvcaller_model)
-                .genotyped_intervals_vcf
+                .genotyped_filtered_segments_vcf
                 .collect{it[1]}
                 .set { gcnvcaller_vcf }
 

subworkflows/local/variant_calling/call_sv_germlinecnvcaller.nf

@@ -6,21 +6,23 @@ include { GATK4_COLLECTREADCOUNTS             } from '../../../modules/nf-core/g
 include { GATK4_DETERMINEGERMLINECONTIGPLOIDY } from '../../../modules/nf-core/gatk4/determinegermlinecontigploidy/main.nf'
 include { GATK4_GERMLINECNVCALLER             } from '../../../modules/nf-core/gatk4/germlinecnvcaller/main.nf'
 include { GATK4_POSTPROCESSGERMLINECNVCALLS   } from '../../../modules/nf-core/gatk4/postprocessgermlinecnvcalls/main.nf'
+include { BCFTOOLS_VIEW                       } from '../../../modules/nf-core/bcftools/view/main'
+include { TABIX_TABIX                         } from '../../../modules/nf-core/tabix/tabix/main'
 
 workflow CALL_SV_GERMLINECNVCALLER {
     take:
         ch_bam_bai             // channel: [mandatory][ val(meta), path(bam), path(bai) ]
         ch_fasta               // channel: [mandatory][ val(meta), path(ch_fasta_no_meta) ]
         ch_fai                 // channel: [mandatory][ val(meta), path(ch_fai) ]
-        ch_readcount_intervals // channel: [mandatory][ val(meta), path(bed), path(tbi) ]
+        ch_readcount_intervals // channel: [mandatory][ path(intervals) ]
         ch_genome_dictionary   // channel: [mandatory][ val(meta), path(ch_dict) ]
         ch_ploidy_model        // channel: [mandatory][ path(ch_ploidy_model) ]
         ch_gcnvcaller_model    // channel: [mandatory][ path(ch_gcnvcaller_model) ]
 
     main:
         ch_versions = Channel.empty()
 
-        input = ch_bam_bai.combine( ch_readcount_intervals.collect{ it[1] } )
+        input = ch_bam_bai.combine( ch_readcount_intervals )
 
         GATK4_COLLECTREADCOUNTS ( input, ch_fasta, ch_fai, ch_genome_dictionary )
 
@@ -47,16 +49,26 @@ workflow CALL_SV_GERMLINECNVCALLER {
 
         GATK4_POSTPROCESSGERMLINECNVCALLS ( ch_postproc_in )
 
+        TABIX_TABIX(GATK4_POSTPROCESSGERMLINECNVCALLS.out.segments)
+        GATK4_POSTPROCESSGERMLINECNVCALLS.out.segments
+            .join(TABIX_TABIX.out.tbi, failOnMismatch:true)
+            .set {ch_segments_in}
+        // Filter out reference only (0/0) segments
+        BCFTOOLS_VIEW (ch_segments_in , [], [], [] )
+
         ch_versions = ch_versions.mix(GATK4_COLLECTREADCOUNTS.out.versions)
         ch_versions = ch_versions.mix(GATK4_DETERMINEGERMLINECONTIGPLOIDY.out.versions)
         ch_versions = ch_versions.mix(GATK4_GERMLINECNVCALLER.out.versions)
         ch_versions = ch_versions.mix(GATK4_POSTPROCESSGERMLINECNVCALLS.out.versions)
+        ch_versions = ch_versions.mix(TABIX_TABIX.out.versions)
+        ch_versions = ch_versions.mix(BCFTOOLS_VIEW.out.versions)
 
     emit:
-        genotyped_intervals_vcf = GATK4_POSTPROCESSGERMLINECNVCALLS.out.intervals  // channel: [ val(meta), path(*.tar.gz) ]
-        genotyped_segments_vcf  = GATK4_POSTPROCESSGERMLINECNVCALLS.out.segments   // channel: [ val(meta), path(*.tar.gz) ]
-        denoised_vcf            = GATK4_POSTPROCESSGERMLINECNVCALLS.out.denoised   // channel: [ val(meta), path(*.tar.gz) ]
-        versions                = ch_versions                                      // channel: [ versions.yml ]
+        genotyped_intervals_vcf          = GATK4_POSTPROCESSGERMLINECNVCALLS.out.intervals  // channel: [ val(meta), path(*.vcf.gz) ]
+        genotyped_segments_vcf           = GATK4_POSTPROCESSGERMLINECNVCALLS.out.segments   // channel: [ val(meta), path(*.vcf.gz) ]
+        genotyped_filtered_segments_vcf  = BCFTOOLS_VIEW.out.vcf                            // channel: [ val(meta), path(*.vcf.gz) ]
+        denoised_vcf                     = GATK4_POSTPROCESSGERMLINECNVCALLS.out.denoised   // channel: [ val(meta), path(*.vcf.gz) ]
+        versions                         = ch_versions                                      // channel: [ versions.yml ]
 }
 
 // This function groups calls with same meta for postprocessing.