Add GRIDSS variant calling

CHANGELOG.md

@@ -9,6 +9,7 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ## [Unreleased]
 ### Added
+- Add GRIDSS2 variant calling
 - Add GRIDSS2 assembly
 - Add GRIDSS2 preprocessing
 - Add supported Nextflow version to `README.md`

config/F16.config

@@ -26,6 +26,17 @@ process {
             }
         }
 
+    withName: call_sSV_GRIDSS {
+        cpus = 8
+        memory = 15.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 16.GB

config/F2.config

@@ -36,7 +36,7 @@ process {
                 }
             }
         }
-    
+
     withName: call_sSV_Manta {
         cpus = 1
         memory = 3.GB
@@ -47,5 +47,4 @@ process {
                 }
             }
         }
-    } 
-
+    }

config/F32.config

@@ -26,6 +26,17 @@ process {
             }
         }
 
+    withName: call_sSV_GRIDSS {
+        cpus = 8
+        memory = 30.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 30.GB

config/F72.config

@@ -26,6 +26,17 @@ process {
             }
         }
 
+    withName: call_sSV_GRIDSS {
+        cpus = 8
+        memory = 30.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 30.GB

config/M64.config

@@ -26,6 +26,17 @@ process {
             }
         }
 
+    withName: call_sSV_GRIDSS {
+        cpus = 8
+        memory = 120.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 120.GB

config/template.config

@@ -16,6 +16,7 @@ params {
     // GRCh38 blacklist - /hot/resource/tool-specific-input/GRIDSS-2.13.2/GRCh38-BI-20160721/ENCFF356LFX.bed
     gridss_blacklist = "/hot/resource/tool-specific-input/GRIDSS-2.13.2/GRCh38-BI-20160721/ENCFF356LFX.bed"
     gridss_reference_fasta = "/hot/resource/tool-specific-input/GRIDSS-2.13.2/GRCh38-BI-20160721/Homo_sapiens_assembly38.fasta"
+    other_jvm_heap = 4.GB
 
     exclusion_file = "/hot/resource/tool-specific-input/Delly/hg38/human.hg38.excl.tsv"
 

main.nf

@@ -55,7 +55,7 @@ include { call_sSV_Delly; filter_sSV_Delly } from './module/delly' addParams(
 include { call_sSV_Manta } from './module/manta' addParams(
     workflow_output_dir: "${params.output_dir_base}/Manta-${params.manta_version}"
     )
-include { preprocess_BAM_GRIDSS; run_assembly_GRIDSS } from './module/gridss' addParams(
+include { preprocess_BAM_GRIDSS; run_assembly_GRIDSS; call_sSV_GRIDSS } from './module/gridss' addParams(
     workflow_output_dir: "${params.output_dir_base}/GRIDSS-${params.gridss_version}"
     )
 include { generate_sha512 as generate_sha512_BCFtools } from './module/sha512' addParams(
@@ -217,6 +217,21 @@ workflow {
             gridss_reference_files,
             params.gridss_blacklist
             )
-        }
 
+        gridss_assembly_dir = run_assembly_GRIDSS.out.gridss_assembly
+            .flatten()
+            .map { parentdir -> parentdir.getParent() }
+            .unique()
+            .collect()
+
+        call_sSV_GRIDSS(
+            input_paired_bams_ch,
+            gridss_preprocess_dir,
+            gridss_assembly_dir,
+            run_assembly_GRIDSS.out.gridss_assembly_bam,
+            params.gridss_reference_fasta,
+            gridss_reference_files,
+            params.gridss_blacklist
+            )
+        }
     }

module/gridss.nf

@@ -14,6 +14,7 @@ process preprocess_BAM_GRIDSS {
     container params.docker_image_gridss
 
     publishDir "${params.workflow_output_dir}/intermediate/${task.process.replace(':', '/')}",
+        enabled: params.save_intermediate_files,
         pattern: "${bam_name}.gridss.working/*",
         mode: "copy",
         saveAs: {
@@ -61,13 +62,15 @@ process run_assembly_GRIDSS {
     container params.docker_image_gridss
 
     publishDir "${params.workflow_output_dir}/intermediate/${task.process.replace(':', '/')}",
+        enabled: params.save_intermediate_files,
         pattern: "${tumor_id}.assembly.bam",
         mode: "copy",
         saveAs: {
             "${output_filename}_${sanitize_string(file(it).getName().replace("${tumor_id}.", ""))}"
             }
 
     publishDir "${params.workflow_output_dir}/intermediate/${task.process.replace(':', '/')}",
+        enabled: params.save_intermediate_files,
         pattern: "${tumor_id}.assembly.bam.gridss.working/*",
         mode: "copy",
         saveAs: {
@@ -88,11 +91,11 @@ process run_assembly_GRIDSS {
 
     output:
         path "${tumor_id}.assembly.bam", emit: gridss_assembly_bam
-        path "${tumor_id}.assembly.bam.gridss.working/*"
+        path "${tumor_id}.assembly.bam.gridss.working/*", emit: gridss_assembly
         path ".command.*"
 
     script:
-        otherjvmheap = 4.GB
+        otherjvmheap = params.other_jvm_heap
         gridss_otherjvmheap = "${otherjvmheap.toGiga()}g"
         gridss_jvmheap = "${(task.memory - otherjvmheap).toGiga()}g"
         gridss_jar = "/usr/local/share/gridss-${params.gridss_version}-1/gridss.jar"
@@ -118,3 +121,70 @@ process run_assembly_GRIDSS {
             ${tumor_bam}
         """
     }
+
+process call_sSV_GRIDSS {
+    container params.docker_image_gridss
+
+    publishDir "${params.workflow_output_dir}/output/",
+        pattern: "${tumor_id}.{vcf,vcf.idx}",
+        mode: "copy",
+        saveAs: {
+            "${output_filename}.${sanitize_string(file(it).getName().replace("${tumor_id}.", ""))}"
+            }
+
+    publishDir "${params.workflow_output_dir}/intermediate/${task.process.replace(':', '/')}",
+        enabled: params.save_intermediate_files,
+        pattern: "${tumor_id}.vcf.gridss.working/*",
+        mode: "copy",
+        saveAs: {
+            "${output_filename}.vcf.gridss.working/${output_filename}.${sanitize_string(file(it).getName().replace("${tumor_id}.", ""))}"
+            }
+
+    publishDir "${params.log_output_dir}/process-log",
+        pattern: ".command.*",
+        mode: "copy",
+        saveAs: { "${task.process.replace(':', '/')}/log${file(it).getName()}" }
+
+    input:
+        tuple(val(tumor_id), path(tumor_bam), path(tumor_bai), path(normal_bam), path(normal_bai))
+        path(gridss_preprocess_dir)
+        path(gridss_assembly_dir)
+        path(gridss_assembly_bam)
+        path(gridss_reference_fasta)
+        path(gridss_reference_files)
+        path(gridss_blacklist)
+
+    output:
+        path "${tumor_id}.vcf", emit: gridss_vcf
+        path "${tumor_id}.vcf.idx", emit: gridss_vcf_idx
+        path "${tumor_id}.vcf.gridss.working/*", emit: gridss_vcf_dir
+        path ".command.*"
+
+    script:
+        otherjvmheap = params.other_jvm_heap
+        gridss_otherjvmheap = "${otherjvmheap.toGiga()}g"
+        gridss_jvmheap = "${(task.memory - otherjvmheap).toGiga()}g"
+        gridss_jar = "/usr/local/share/gridss-${params.gridss_version}-1/gridss.jar"
+        output_filename = generate_standard_filename(
+            "GRIDSS2-${params.gridss_version}",
+            params.dataset_id,
+            tumor_id,
+            [:]
+            )
+
+        """
+        set -euo pipefail
+        gridss \
+            -r ${gridss_reference_fasta} \
+            -j ${gridss_jar} \
+            -s call \
+            -t ${task.cpus} \
+            --jvmheap ${gridss_jvmheap} \
+            --otherjvmheap ${gridss_otherjvmheap} \
+            -b ${gridss_blacklist} \
+            -a ${gridss_assembly_bam} \
+            --output ${tumor_id}.vcf \
+            ${normal_bam} \
+            ${tumor_bam}
+        """
+    }