Make sigrap (HRDetect, MutationalPatterns) individual process

conf/modules.config

@@ -126,7 +126,31 @@ process {
         publishDir = [
             path: { "${params.outdir}" },
             mode: params.publish_dir_mode,
-            saveAs: { filename -> filename.equals('versions.yml') ? null : "${meta.key}/${filename}" },
+            saveAs: { filename -> filename.equals('versions.yml') ? null : "${meta.key}/${filename}" }
+        ]
+    }
+
+
+    withName: 'SIGRAP_HRDETECT' {
+        publishDir = [
+            path: { "${params.outdir}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : "${meta.key}/sigrap/hrdetect/${filename}" }
+        ]
+    }
+
+    withName: 'SIGRAP_MUTPAT' {
+        publishDir = [
+            path: { "${params.outdir}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> 
+                if (filename.equals('versions.yml')) {
+                    return null
+                } else {
+                    println "SIGRAP_MUTPAT - meta.key: '${meta.key}', filename: '${filename}'"
+                    return "${meta.key}/sigrap/${filename}"
+                }
+            }
         ]
     }
 

modules/local/bolt/other/cancer_report/main.nf

@@ -5,7 +5,7 @@ process BOLT_OTHER_CANCER_REPORT {
     container 'ghcr.io/umccr/bolt:0.2.14-gpgr'
 
     input:
-    tuple val(meta), path(smlv_somatic_vcf), path(smlv_somatic_bcftools_stats), path(smlv_somatic_counts_process), path(sv_somatic_tsv), path(sv_somatic_vcf), path(cnv_somatic_tsv), path(af_global), path(af_keygenes), path(purple_baf_plot), path(purple_dir), path(virusbreakend_dir), path(dragen_hrd)
+    tuple val(meta), path(smlv_somatic_vcf), path(smlv_somatic_bcftools_stats), path(smlv_somatic_counts_process), path(sv_somatic_tsv), path(sv_somatic_vcf), path(cnv_somatic_tsv), path(af_global), path(af_keygenes), path(purple_baf_plot), path(purple_dir), path(virusbreakend_dir), path(dragen_hrd), path(smlv_somatic_mutpat), path(smlv_somatic_hrdetect), path(smlv_somatic_chord)
     path somatic_driver_panel
     path oncokb_genes
 
@@ -33,6 +33,10 @@ process BOLT_OTHER_CANCER_REPORT {
         --smlv_somatic_bcftools_stats_fp \$(pwd)/${smlv_somatic_bcftools_stats} \\
         --smlv_somatic_counts_process_fp \$(pwd)/${smlv_somatic_counts_process} \\
         \\
+        --mutpat_dir \$(pwd)/${smlv_somatic_mutpat} \\
+        --hrdetect_file \$(pwd)/${smlv_somatic_hrdetect} \\
+        --chord_file \$(pwd)/${smlv_somatic_chord} \\
+        \\
         --sv_somatic_tsv_fp \$(pwd)/${sv_somatic_tsv} \\
         --sv_somatic_vcf_fp \$(pwd)/${sv_somatic_vcf} \\
         --cnv_somatic_tsv_fp \$(pwd)/${cnv_somatic_tsv} \\

modules/local/sigrap/chord/main.nf

@@ -0,0 +1,38 @@
+process SIGRAP_CHORD {
+    tag "${meta.id}"
+    label 'process_low'
+
+    container 'ghcr.io/umccr/sigrap:0.2.0'
+
+    input:
+    tuple val(meta), path(chord_prediction_tsv)
+
+    output:
+    tuple val(meta), path('chord.json.gz')                  , emit: chord_json
+    path 'versions.yml'                                       , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+
+    """
+    sigrap.R chord \\
+        --sample ${meta.subject_id} \\
+        --chord ${chord_prediction_tsv} \\
+        --out chord.json.gz
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        sigrap: \$(sigrap.R --version | sed 's/^.*version //')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    mkdir -p output/
+    touch chord.json.gz
+    echo -e '${task.process}:\\n  stub: noversions\\n' > versions.yml
+    """
+}

modules/local/sigrap/hrdetect/main.nf

@@ -0,0 +1,39 @@
+process SIGRAP_HRDETECT {
+    tag "${meta.id}"
+    label 'process_low'
+
+    container 'ghcr.io/umccr/sigrap:0.2.0'
+
+    input:
+    tuple val(meta), path(smlv_somatic_vcf), path(sv_somatic_vcf), path(cnv_somatic_tsv)
+
+    output:
+    tuple val(meta), path('hrdetect.json.gz')                 , emit: hrdetect_json
+    path 'versions.yml'                                       , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+
+    """
+    sigrap.R hrdetect \\
+        --sample ${meta.subject_id} \\
+        --snv ${smlv_somatic_vcf} \\
+        --sv ${sv_somatic_vcf} \\
+        --cnv ${cnv_somatic_tsv} \\
+        --out  hrdetect.json.gz
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        sigrap: \$(sigrap.R --version | sed 's/^.*version //')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    touch hrdetect.json.gz
+    echo -e '${task.process}:\\n  stub: noversions\\n' > versions.yml
+    """
+}

modules/local/sigrap/mutpat/main.nf

@@ -0,0 +1,40 @@
+process SIGRAP_MUTPAT {
+    tag "${meta.id}"
+    label 'process_low'
+
+    container 'ghcr.io/umccr/sigrap:0.2.0'
+
+    input:
+    tuple val(meta), path(smlv_somatic_vcf)
+
+    output:
+    tuple val(meta), path('mutpat/')                   , emit: mutpat_output
+    path 'versions.yml'                                       , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+
+    """
+    sigrap.R mutpat \\
+        --sample ${meta.subject_id} \\
+        --snv ${smlv_somatic_vcf} \\
+        --rainfall \\
+        --strand-bias \\
+        --out mutpat/
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        sigrap: \$(sigrap.R --version | sed 's/^.*version //')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    mkdir -p sigrap/mutpat/
+    touch sigrap/mutpat/stub_output
+    echo -e '${task.process}:\\n  stub: noversions\\n' > versions.yml
+    """
+}

subworkflows/local/prepare_input.nf

@@ -90,6 +90,13 @@ workflow PREPARE_INPUT {
             return [meta, "${base}/virusbreakend/"]
         }
 
+        // CHORD: homologous recombination deficiency prediction
+        // channel: [ meta, chord_prediction_tsv ]
+        ch_chord = ch_metas.map { meta ->
+            def base = file(meta.oncoanalyser_dir).toUriString()
+            return [meta, "${base}/chord/${meta.tumor_id}.chord.prediction.tsv"]
+        }
+
         // HRD: homologous recombination deficiency scores
         // channel: [ meta, hrdscore_csv ]
         ch_input_hrd = ch_metas.map { meta ->
@@ -120,6 +127,7 @@ workflow PREPARE_INPUT {
         cobalt           = ch_cobalt                  // channel: [ meta, cobalt_dir ]
         sage_somatic     = ch_sage_somatic            // channel: [ meta, sage_somatic_vcf, sage_somatic_tbi ]
         virusbreakend    = ch_virusbreakend           // channel: [ meta, virusbreakend_dir ]
+        chord            = ch_chord                   // channel: [ meta, chord_prediction_tsv ]
         call_inputs      = ch_call_inputs             // channel: [ meta_esvee, esvee_ref_depth_vcf, esvee_prep_dir ]
 
         // DRAGEN channels

workflows/sash.nf

@@ -42,13 +42,15 @@ include { BOLT_SMLV_GERMLINE_PREPARE } from '../modules/local/bolt/smlv_germline
 include { BOLT_SMLV_GERMLINE_REPORT  } from '../modules/local/bolt/smlv_germline/report/main'
 include { BOLT_SMLV_SOMATIC_ANNOTATE } from '../modules/local/bolt/smlv_somatic/annotate/main'
 include { BOLT_SMLV_SOMATIC_FILTER   } from '../modules/local/bolt/smlv_somatic/filter/main'
-include { BOLT_SMLV_SOMATIC_REPORT   } from '../modules/local/bolt/smlv_somatic/report/main'
 include { BOLT_SMLV_SOMATIC_RESCUE   } from '../modules/local/bolt/smlv_somatic/rescue/main'
+include { BOLT_SMLV_SOMATIC_REPORT   } from '../modules/local/bolt/smlv_somatic/report/main'
 include { BOLT_SV_SOMATIC_ANNOTATE   } from '../modules/local/bolt/sv_somatic/annotate/main'
 include { BOLT_SV_SOMATIC_PRIORITISE } from '../modules/local/bolt/sv_somatic/prioritise/main'
-include { ESVEE_CALL                 } from '../modules/local/esvee/call/main'
 include { PAVE_SOMATIC               } from '../modules/local/pave/somatic/main'
+include { SIGRAP_HRDETECT            } from '../modules/local/sigrap/hrdetect/main'
+include { SIGRAP_MUTPAT              } from '../modules/local/sigrap/mutpat/main'
 
+include { ESVEE_CALL                 } from '../modules/local/esvee/call/main'
 include { LINX_ANNOTATION            } from '../subworkflows/local/linx_annotation'
 include { LINX_PLOTTING              } from '../subworkflows/local/linx_plotting'
 include { PREPARE_INPUT              } from '../subworkflows/local/prepare_input'
@@ -93,6 +95,7 @@ workflow SASH {
     ch_sage_somatic        = PREPARE_INPUT.out.sage_somatic      // channel: [ meta, sage_somatic_vcf, sage_somatic_tbi ]
     ch_virusbreakend       = PREPARE_INPUT.out.virusbreakend     // channel: [ meta, virusbreakend_dir ]
     ch_call_inputs         = PREPARE_INPUT.out.call_inputs       // channel: [ meta_esvee, esvee_ref_depth_vcf, esvee_prep_dir ]
+    ch_chord               = PREPARE_INPUT.out.chord             // channel: [ meta, chord_prediction_tsv ]
 
     // DRAGEN inputs
     ch_input_hrd           = PREPARE_INPUT.out.hrd               // channel: [ meta, hrdscore_csv ]
@@ -120,8 +123,7 @@ workflow SASH {
     // Somatic small variants
     //
 
-
-
+    // Prepare rescue inputs with meta transformation
     // channel: [ meta_bolt, dragen_somatic_vcf, dragen_somatic_tbi, sage_somatic_vcf, sage_somatic_tbi ]
     ch_smlv_somatic_rescue_inputs = WorkflowSash.groupByMeta(
         ch_input_vcf_somatic,
@@ -152,6 +154,7 @@ workflow SASH {
         umccr_data.annotations_dir,
         misc_data.pon_dir,
         misc_data.pcgr_dir,
+        misc_data.vep_dir
     )
 
     ch_versions = ch_versions.mix(BOLT_SMLV_SOMATIC_ANNOTATE.out.versions)
@@ -162,6 +165,7 @@ workflow SASH {
 
     ch_versions = ch_versions.mix(BOLT_SMLV_SOMATIC_FILTER.out.versions)
 
+    // Restore meta and create clean outputs
     // channel: [ meta, smlv_somatic_vcf ]
     ch_smlv_somatic_out = WorkflowSash.restoreMeta(BOLT_SMLV_SOMATIC_FILTER.out.vcf, ch_inputs)
         .map { meta, vcf, tbi -> [meta, vcf] }
@@ -180,6 +184,8 @@ workflow SASH {
         umccr_data.ensembl_data_resources,
     )
 
+    ch_versions = ch_versions.mix(PAVE_SOMATIC.out.versions)
+
     // channel: [ meta, pave_somatic_vcf ]
     ch_pave_somatic_out = WorkflowSash.restoreMeta(PAVE_SOMATIC.out.vcf, ch_inputs)
 
@@ -218,7 +224,6 @@ workflow SASH {
 
     //
     // Somatic structural variants
-
     //
 
     ESVEE_CALL(
@@ -429,11 +434,60 @@ workflow SASH {
 
 
 
+    //
+    // Sigrap
+    //
+
+    // channel: [ meta_sigrap, smlv_somatic_vcf, sv_somatic_vcf, cnv_somatic_tsv ]
+    ch_sigrap_hrdetect_inputs = WorkflowSash.groupByMeta(
+        ch_smlv_somatic_out,
+        ch_sv_somatic_sv_vcf_out,
+        ch_sv_somatic_cnv_tsv_out,
+    )
+        .map { meta, smlv_vcf, sv_vcf, cnv_tsv -> 
+            def meta_sigrap = [
+                key: meta.id,
+                id: meta.id,
+                tumor_id: meta.tumor_id,
+            ]
+            return [meta_sigrap, smlv_vcf, sv_vcf, cnv_tsv] 
+        }
+
+    SIGRAP_HRDETECT(
+        ch_sigrap_hrdetect_inputs
+    )
+
+    // channel: [ meta, hrdetect_json ]
+    ch_sigrap_hrdetect = SIGRAP_HRDETECT.out.hrdetect_json
+    ch_versions = ch_versions.mix(SIGRAP_HRDETECT.out.versions)
+
+    // channel: [ meta_sigrap, smlv_somatic_vcf ]
+    ch_sigrap_mutpat_inputs = ch_smlv_somatic_out.map { meta, vcf -> 
+        def meta_sigrap = [
+            key: meta.id,
+            id: meta.id,
+            tumor_id: meta.tumor_id,
+        ]
+        return [meta_sigrap, vcf] 
+    }
+
+    SIGRAP_MUTPAT(
+        ch_sigrap_mutpat_inputs
+    )
+
+    // channel: [ meta, mutpat_output ]
+    ch_sigrap_mutpat = SIGRAP_MUTPAT.out.mutpat_output
+    ch_versions = ch_versions.mix(SIGRAP_MUTPAT.out.versions)
+
+
+
+
+
     //
     // Generate the cancer report
     //
 
-    // channel: [ meta_bolt, smlv_somatic_vcf, smlv_somatic_bcftools_stats, smlv_somatic_counts_process, sv_tsv, sv_vcf, cnv_tsv, af_global, af_keygenes, purple_baf_circos_plot, purple_dir, virusbreakend_dir, dragen_hrd ]
+    // channel: [ meta_bolt, smlv_somatic_vcf, smlv_somatic_bcftools_stats, smlv_somatic_counts_process, sv_tsv, sv_vcf, cnv_tsv, af_global, af_keygenes, purple_baf_circos_plot, purple_dir, virusbreakend_dir, dragen_hrd, mutpat, hrdetect, chord ]
     ch_cancer_report_inputs = WorkflowSash.groupByMeta(
         ch_smlv_somatic_out,
         ch_smlv_somatic_report_stats_out,
@@ -447,6 +501,9 @@ workflow SASH {
         PURPLE_CALLING.out.purple_dir,
         ch_virusbreakend,
         ch_input_hrd,
+        ch_sigrap_mutpat,
+        ch_sigrap_hrdetect,
+        ch_chord
     )
         .map {
             def meta = it[0]