CNVkit/tasks/batch.wdl

task batch {
    String sample_id
    Array[File] tumor_bam
    Array[File] normal_bam
    Array[File] bam_index
    File bed
    File faidx
    File fasta
    File? ref_flat
    File access_bed
    File? reference
    String method
    String segment_method
    String docker
    String cluster_config
    String disk_size

    String annotate_opt = if (ref_flat != "") then "--annotate ref_flat.txt" else ""
    String reference_opt = if (reference != "") then "--reference my_reference.cnn" else ""
    String access_opt = if (method == "amplicon") then "--access target.bed" else "--access access-mappable.bed"

    command <<<
        set -o pipefail
        set -e
        nt=$(nproc)

        mkdir -p /cromwell_root/tmp/cnvkit
        cp -r ${sep=" " bam_index} /cromwell_root/tmp/cnvkit

        # must exist parameters
        cp ${fasta} /cromwell_root/tmp/cnvkit/hg38.fa
        cp ${faidx} /cromwell_root/tmp/cnvkit/hg38.fai
        cp ${bed} /cromwell_root/tmp/cnvkit/target.bed
        cp ${access_bed} /cromwell_root/tmp/cnvkit/access-mappable.bed

        # optional parameters
        if [ ${ref_flat} != "" ]; then cp ${ref_flat} /cromwell_root/tmp/cnvkit/ref_flat.txt; fi
        if [ ${reference} != "" ]; then cp ${reference} /cromwell_root/tmp/cnvkit/my_reference.cnn; fi
        
        cd /cromwell_root/tmp/cnvkit
        mkdir results
        cnvkit.py batch -p $nt ${sep=' ' tumor_bam} --normal ${sep=' ' normal_bam} \
        --method ${method} --segment-method ${segment_method} \
        --targets target.bed ${access_opt} ${annotate_opt} \
        --fasta hg38.fa ${reference_opt} \
        --output-reference ${sample_id}.reference.cnn \
        --output-dir ./results/ \
        --drop-low-coverage --diagram --scatter
        
        cp ${sample_id}.reference.cnn ~/${sample_id}.reference.cnn
    >>>

    runtime { 
        docker: docker
        cluster: cluster_config
        systemDisk: "cloud_ssd 40"
        dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
    }

    output {
        File reference_cnn = "${sample_id}.reference.cnn"
        Array[File] results = glob("/cromwell_root/tmp/cnvkit/results/*")
    }
}