Add GRIDSS Assembly (#179)

* update gridss related changes * update gridss preprocess into configs and add assembly process * update gridss preprocess a and add assembly * update preprocess coin main.nf and add assembly * fix emitting of preprocess to assembly * fifix intermediate publish dir * update memory for assembly * Update CHANGELOG.md * uncomment pipeval * parse preprocess dir paths and use as input to assembly * remove redundant view() * add otherjnmheap parameter to assembly and adjust jvmheap * optimize F32 memory for assembly * optimize F72 memory for assembly * optimize F72 CPU for assembly * set otherjvmheap value to 4GB and subtract it from task.memory to set jvmheap * remove redundant collect() * publish assembly BAM into intermediate dir --------- Co-authored-by: Mootor <mmootor@ip-0A125937.rhxrlfvjyzbupc03cc22jkch3c.xx.internal.cloudapp.net>
uclahs-cds · Oct 17, 2024 · 9690887 · 9690887
1 parent 253dd16
commit 9690887
Show file tree

Hide file tree

Showing 9 changed files with 129 additions and 1 deletion.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -9,6 +9,7 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ## [Unreleased]
 ### Added
+- Add GRIDSS2 assembly
 - Add GRIDSS2 preprocessing
 - Add supported Nextflow version to `README.md`
 - Add PlantUML diagram

diff --git a/config/F16.config b/config/F16.config
@@ -15,6 +15,17 @@ process {
             }
         }
 
+    withName: run_assembly_GRIDSS {
+        cpus = 8
+        memory = 15.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 16.GB

diff --git a/config/F32.config b/config/F32.config
@@ -15,6 +15,17 @@ process {
             }
         }
 
+    withName: run_assembly_GRIDSS {
+        cpus = 8
+        memory = 25.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 30.GB

diff --git a/config/F72.config b/config/F72.config
@@ -15,6 +15,17 @@ process {
             }
         }
 
+    withName: run_assembly_GRIDSS {
+        cpus = 8
+        memory = 35.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 30.GB

diff --git a/config/M64.config b/config/M64.config
@@ -15,6 +15,17 @@ process {
             }
         }
 
+    withName: run_assembly_GRIDSS {
+        cpus = 8
+        memory = 120.GB
+        retry_strategy {
+            memory {
+                strategy = 'exponential'
+                operand = 2
+                }
+            }
+        }
+
     withName: call_sSV_Delly {
         cpus = 1
         memory = 120.GB

diff --git a/config/schema.yaml b/config/schema.yaml
@@ -13,6 +13,11 @@ gridss_reference_fasta:
   mode: 'r'
   required: true
   help: 'Absolute path to GRIDSS2 reference FASTA file'
+gridss_blacklist:
+  type: 'Path'
+  mode: 'r'
+  required: true
+  help: 'Absolute path to GRIDSS2 blacklist BED file'
 exclusion_file:
   type: 'Path'
   mode: 'r'

diff --git a/config/template.config b/config/template.config
@@ -12,6 +12,9 @@ params {
     blcds_registered_dataset = false
 
     reference_fasta = "/hot/ref/reference/GRCh38-BI-20160721/Homo_sapiens_assembly38.fasta"
+    // GRCh37 blacklist - /hot/ref/tool-specific-input/GRIDSS-2.13.2/GRCh37-EBI-hs37d5/ENCFF001TDO.bed
+    // GRCh38 blacklist - /hot/ref/tool-specific-input/GRIDSS-2.13.2/GRCh38-BI-20160721/ENCFF356LFX.bed
+    gridss_blacklist = "/hot/ref/tool-specific-input/GRIDSS-2.13.2/GRCh38-BI-20160721/ENCFF356LFX.bed"
     gridss_reference_fasta = "/hot/ref/tool-specific-input/GRIDSS-2.13.2/GRCh38-BI-20160721/Homo_sapiens_assembly38.fasta"
 
     exclusion_file = "/hot/ref/tool-specific-input/Delly/hg38/human.hg38.excl.tsv"

diff --git a/main.nf b/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 } from './module/gridss' addParams(
+include { preprocess_BAM_GRIDSS; run_assembly_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(
@@ -204,6 +204,19 @@ workflow {
             params.gridss_reference_fasta,
             gridss_reference_files
             )
+        gridss_preprocess_dir = preprocess_BAM_GRIDSS.out.gridss_preprocess
+            .flatten()
+            .map { parentdir -> parentdir.getParent() }
+            .unique()
+            .collect()
+
+        run_assembly_GRIDSS(
+            input_paired_bams_ch,
+            gridss_preprocess_dir,
+            params.gridss_reference_fasta,
+            gridss_reference_files,
+            params.gridss_blacklist
+            )
         }
 
     }
diff --git a/module/gridss.nf b/module/gridss.nf
@@ -56,3 +56,65 @@ process preprocess_BAM_GRIDSS {
             ${sample_bam}
         """
     }
+
+process run_assembly_GRIDSS {
+    container params.docker_image_gridss
+
+    publishDir "${params.workflow_output_dir}/intermediate/${task.process.replace(':', '/')}",
+        pattern: "${tumor_id}.assembly.bam",
+        mode: "copy",
+        saveAs: {
+            "${output_filename}_${sanitize_string(file(it).getName().replace("${tumor_id}.", ""))}"
+            }
+
+    publishDir "${params.workflow_output_dir}/intermediate/${task.process.replace(':', '/')}",
+        pattern: "${tumor_id}.assembly.bam.gridss.working/*",
+        mode: "copy",
+        saveAs: {
+            "${output_filename}.assembly.bam.gridss.working/${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${file(it).getName()}" }
+
+    input:
+        tuple(val(tumor_id), path(tumor_bam), path(tumor_bai), path(normal_bam), path(normal_bai))
+        path(gridss_preprocess_dir)
+        path(gridss_reference_fasta)
+        path(gridss_reference_files)
+        path(gridss_blacklist)
+
+    output:
+        path "${tumor_id}.assembly.bam", emit: gridss_assembly_bam
+        path "${tumor_id}.assembly.bam.gridss.working/*"
+        path ".command.*"
+
+    script:
+        otherjvmheap = 4.GB
+        gridss_otherjvmheap = "${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 assemble \
+            -t ${task.cpus} \
+            --jvmheap ${gridss_jvmheap} \
+            --otherjvmheap ${gridss_otherjvmheap} \
+            -b ${gridss_blacklist} \
+            -a ${tumor_id}.assembly.bam \
+            ${normal_bam} \
+            ${tumor_bam}
+        """
+    }