Clincnv

conf/modules.config

@@ -36,6 +36,17 @@ process {
         ]
     }
 
+    withName: 'BEDCOVERAGE' {
+        ext.args = {[
+            '-min_mapq 3',
+            '-clear'
+            ].join(' ').trim()}
+    }
+
+    withName: 'COUNT_MERGE' {
+        ext.args = { "-n 4" }
+    }
+
     withName: 'ENSEMBLVEP_VEP' {
         ext.prefix = {"${meta.id}.vep"}
         ext.args = {[
@@ -52,6 +63,21 @@ process {
         ].join(' ').trim()}
     }
 
+    withName: 'GERMLINE' {
+        ext.args = {[
+            '--minimumNumOfElemsInCluster 20',
+            '--maxNumGermCNVs 2000',
+            '--maxNumIter 5',
+            '--numberOfThreads 10',
+            '--scoreG 60',
+            '--lengthG 0',
+            '--hg38',
+            '--reanalyseCohort FALSE',
+            '--visulizationIGV F',
+            '--noPlot T'
+        ].join(' ').trim()}
+    }
+
     withName: 'TABIX_TABIX' {
         publishDir = [
             path: {"$params.outdir/exomedepth/cnv_call"},

conf/test.config

@@ -31,9 +31,11 @@ params {
     // Parameters
     roi_auto = "https://raw.githubusercontent.com/nf-cmgg/test-datasets/exomecnv/data/genomics/homo_sapiens/genome/roi/Homo_sapiens.GRCh38.105.chr21_protein_coding_basic_sorted_merged_autosomal.bed"
     roi_chrx = "https://raw.githubusercontent.com/nf-cmgg/test-datasets/exomecnv/data/genomics/homo_sapiens/genome/roi/Homo_sapiens.GRCh38.105.chrX_protein_coding_basic_sorted_merged.bed"
+    roi_merged = "https://raw.githubusercontent.com/nf-cmgg/test-datasets/exomecnv/data/genomics/homo_sapiens/genome/roi/Homo_sapiens.GRCh38.105.chr21X_protein_coding_basic_sorted_merged_autosomal.bed"
 
     yamlconfig = "${projectDir}/assets/exomedepth.yaml"
     exomedepth = false
+    clincnv = false
 
 
     // VEP options
@@ -42,6 +44,5 @@ params {
     vep_assembly        = "GRCh38"
     vep_cache_version   = 105
     species             = "homo_sapiens"
-
 }
 

modules/local/clincnv/bedannotategc/environment.yml

@@ -0,0 +1,7 @@
+name: ngs-env
+channels:
+  - conda-forge
+  - bioconda
+  - defaults
+dependencies:
+  - ngs-bits=2023_02

modules/local/clincnv/bedannotategc/main.nf

@@ -0,0 +1,44 @@
+process BEDANNOTATEGC {
+
+    conda "${moduleDir}/environment.yml"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/ngs-bits:2023_02--py311ha0b7adc_2' :
+        'biocontainers/ngs-bits:2023_02--py311ha0b7adc_2' }"
+
+    publishDir "$params.outdir/clincnv", mode: 'copy'
+
+    input:
+    tuple val(meta), path(exon_target)
+    tuple val(meta2), path(fasta)
+    tuple val(meta3), path(fai)
+
+    output:
+    path("*.bed"), emit: annotatedbed
+    path "versions.yml", emit:versions
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    BedAnnotateGC \\
+        -in $exon_target \\
+        -out GC_Annotated_$exon_target \\
+        -ref $fasta \\
+
+    awk -F'\\t' '{OFS="\\t"; print \$1, \$2, \$3, \$5, \$4}' GC_Annotated_$exon_target > tmp && mv tmp GC_Annotated_$exon_target
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        BedAnnotateGC: \$(BedAnnotateGC --version | sed 's/BedAnnotateGC //g')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    touch GC_Annotated.bed
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        BedCoverage: \$(BedCoverage --version | sed 's/BedCoverage //g')
+    END_VERSIONS
+    """
+}

modules/local/clincnv/bedcoverage/environment.yml

@@ -0,0 +1,7 @@
+name: ngs-env
+channels:
+  - conda-forge
+  - bioconda
+  - defaults
+dependencies:
+  - ngs-bits=2023_02

modules/local/clincnv/bedcoverage/main.nf

@@ -0,0 +1,48 @@
+process BEDCOVERAGE {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda "${moduleDir}/environment.yml"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/ngs-bits:2023_02--py311ha0b7adc_2' :
+        'biocontainers/ngs-bits:2023_02--py311ha0b7adc_2' }"
+
+    publishDir "$params.outdir/clincnv/counts/$meta.pool", mode: 'copy'
+
+    input:
+    tuple val(meta), path(bam), path(bai)
+    tuple val(meta2), path(exon_target)
+
+    output:
+    tuple val(meta), path("*.cov"), emit: counts
+    path "versions.yml", emit:versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    BedCoverage \\
+        -bam $bam \\
+        -in $exon_target \\
+        $args \\
+        -threads $task.cpus \\
+        -out ${meta.id}.cov
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        BedCoverage: \$(BedCoverage --version | sed 's/BedCoverage //g')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    touch ${meta.id}.cov
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        BedCoverage: \$(BedCoverage --version | sed 's/BedCoverage //g')
+    END_VERSIONS
+    """
+}

modules/local/clincnv/germline/environment.yml

@@ -0,0 +1,20 @@
+name: clincnv
+channels:
+  - bioconda
+  - conda-forge
+  - defaults
+dependencies:
+  - r-base=4.2.3
+  - r-rcpp=1.0.11
+  - r-optparse=1.7.5
+  - r-robustbase=0.95_1
+  - r-data.table=1.14.8
+  - r-foreach=1.5.2
+  - r-doparallel=1.0.17
+  - r-mclust=6.0.0
+  - r-r.utils=2.12.2
+  - r-rcolorbrewer=1.1_3
+  - r-party=1.3_13
+  - r-dbscan=1.1_11
+  - r-umap=0.2.10.0
+  - r-mass=7.3_60

modules/local/clincnv/germline/main.nf

@@ -0,0 +1,51 @@
+process GERMLINE {
+    tag "$sample"
+
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/cmgg/clincnv:1.18.3':
+        'cmgg/clincnv:1.18.3' }"
+    conda "${moduleDir}/environment.yml"
+
+    publishDir "$params.outdir/clincnv/calling", mode: 'copy'
+
+    input:
+    tuple val(meta), val(sample), path(file)
+    path(bed)
+
+    output:
+    tuple val(meta), val(sample), path("*.tsv"), emit: tsv
+    path "versions.yml", emit: versions
+
+    script:
+    def publishDir = "$params.outdir/clincnv/calling/"
+    def bedDir = "$params.outdir/clincnv"
+    def fileDir = "$params.outdir/clincnv/counts/$meta.id"
+    def args = task.ext.args ?: ''
+    def VERSION = '1.19.0'
+    """
+    clinCNV.R \\
+        --normal ${fileDir}/${file} \\
+        --normalSample=${sample} \\
+        --bed ${bedDir}/${bed} \\
+        --out ${publishDir} \\
+        $args
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ClinCNV: ${VERSION}
+        R: \$(Rscript --version | sed 's/R scripting front-end //g')
+    END_VERSIONS
+    """
+
+    stub:
+    def VERSION = '1.19.0'
+    """
+    touch ${sample}.ready_CNVS.tsv
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ClinCNV: ${VERSION}
+        R: \$(Rscript --version | sed 's/R scripting front-end //g')
+    END_VERSIONS
+    """
+}

modules/local/clincnv/merge_coverage/environment.yml

@@ -0,0 +1,20 @@
+name: clincnv
+channels:
+  - bioconda
+  - conda-forge
+  - defaults
+dependencies:
+  - r-base=4.2.3
+  - r-rcpp=1.0.11
+  - r-optparse=1.7.5
+  - r-robustbase=0.95_1
+  - r-data.table=1.14.8
+  - r-foreach=1.5.2
+  - r-doparallel=1.0.17
+  - r-mclust=6.0.0
+  - r-r.utils=2.12.2
+  - r-rcolorbrewer=1.1_3
+  - r-party=1.3_13
+  - r-dbscan=1.1_11
+  - r-umap=0.2.10.0
+  - r-mass=7.3_60

modules/local/clincnv/merge_coverage/main.nf

@@ -0,0 +1,48 @@
+// MERGE COUNT FILES
+process COVERAGE_MERGE {
+    tag "$meta.id"
+
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/cmgg/clincnv:1.18.3':
+        'cmgg/clincnv:1.18.3' }"
+    conda "${moduleDir}/environment.yml"
+
+    publishDir "$params.outdir/clincnv/counts/$meta.id", mode: 'copy'
+
+    input:
+    tuple val(meta), val(samples), path(file)
+
+    output:
+    tuple val(meta), val(samples), path("*.txt"), emit: merge
+    path "versions.yml", emit:versions
+
+    script:
+    def prefix = task.ext.prefix ?: "${meta.id}.normal"
+    def args = task.ext.args ?: ''
+    def VERSION = '1.19.0'
+    """
+    mergeFilesFromFolderDT.R \\
+        -i ${file} \\
+        -o ${prefix}.txt \\
+        $args
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ClinCNV: ${VERSION}
+        R: \$(Rscript --version | sed 's/R scripting front-end //g')
+    END_VERSIONS
+    """
+
+    stub:
+    prefix = task.ext.prefix ?: "${meta.id}"
+    def VERSION = '1.19.0'
+    """
+    touch ${prefix}.txt
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ClinCNV: ${VERSION}
+        R: \$(Rscript --version | sed 's/R scripting front-end //g')
+    END_VERSIONS
+    """
+}

modules/local/clincnv/pathfile/main.nf

@@ -0,0 +1,23 @@
+process PATHFILE {
+    tag "$meta.id"
+
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/coreutils:9.3' :
+        'biocontainers/coreutils:9.3' }"
+
+    publishDir "$params.outdir/clincnv/counts/$meta.id", mode: 'copy'
+
+    input:
+    tuple val(meta), val(samples), path(files)
+
+    output:
+    tuple val(meta), val(samples), path("*.txt"), emit: counts
+
+    script:
+    def prefix = task.ext.prefix ?: "${meta.id}_paths"
+    def publishDir = "$params.outdir/clincnv/counts/$meta.id"
+    def fileList = files.collect(file -> "${publishDir}/${file}").join('\n')
+    """
+    echo -e "${fileList}" > ${prefix}.txt
+    """
+}

nextflow.config

@@ -19,7 +19,6 @@ params {
     igenomes_base              = null //'s3://ngi-igenomes/igenomes/'
     igenomes_ignore            = true
 
-
     // MultiQC options
     multiqc_config             = null
     multiqc_title              = null
@@ -72,7 +71,9 @@ params {
     annotate = null
     vep_assembly = "GRCh38"
 
-
+    // CLINCNV Parameters
+    roi_merged  = null
+    clincnv     = false
 }
 
 // Load base.config by default for all pipelines

nextflow_schema.json

@@ -44,6 +44,10 @@
                 "exomedepth": {
                     "type": "boolean",
                     "description": "Option to execute the ExomeDepth tool"
+                },
+                "clincnv": {
+                    "type": "boolean",
+                    "description": "Option to execute the ClinCNV tool"
                 }
             }
         },
@@ -92,9 +96,13 @@
                     "default": "/conf/",
                     "description": "The config base path for the cmgg configs",
                     "format": "directory-path"
+                },
+                "roi_merged": {
+                    "type": "string",
+                    "description": "Path to the default ROI (regions of interest) BED file of autosomal and chrX merged regions to be used for CNV analysis"
                 }
             },
-            "required": ["fasta", "fai"]
+            "required": ["fasta", "fai", "roi_merged"]
         },
         "institutional_config_options": {
             "title": "Institutional config options",