variant-calling/tasks/pindel.wdl

task pindel {
  
  String sample_id
  File bam
  File bam_index
  File ref_dir
  String fasta

  String docker
  String cluster
  String disk_size

  command <<<
	
    set -o pipefail
    set -e

    mkdir ./pindel_result/
    mkdir ./input

    cp ${bam} ./input

    bam_file_name=`echo ${bam}|awk -F "/" '{print $NF}'`
    
    samtools index -@ 4 ./input/$bam_file_name
    
    java "-Xmx16G" -jar /software/picard/picard.jar  CollectInsertSizeMetrics \
      -H ./pindel_result/${sample_id}_picard.pdf \
      -I ./input/$bam_file_name \
      -O ./pindel_result/${sample_id}_picard.txt

    mean_insert_size=`cat pindel_result/${sample_id}_picard.txt|sed -n '8p'|cut -f 6|cut -d . -f 1`
    bam_file_name=`echo ${bam}|awk -F "/" '{print $NF}'`

    echo -e "./input/$bam_file_name\t$mean_insert_size\t${sample_id}" > ${sample_id}_config.txt

    pindel -i ${sample_id}_config.txt \
      -f $${ref_dir}/${fasta} \
      -o ./pindel_result/${sample_id} \
      -c all \
      -T 4 \
      -x 4 \
      -l \
      -B 0 \
      -M 3 \
      -J /software/picard/hg38_ucsc_centromere.bed

    grep "ChrID" pindel_result/${sample_id}_SI > pindel_result/${sample_id}_all_indel
    grep "ChrID" pindel_result/${sample_id}_D >> pindel_result/${sample_id}_all_indel

    awk -v chrID="chr1" '$8==chrID {print}' pindel_result/${sample_id}_all_indel > pindel_result/${sample_id}_indel
    
    for i in `seq 2 22`  X Y M
      do
        awk -v chrID=chr$i '$8==chrID {print}' pindel_result/${sample_id}_all_indel >> pindel_result/${sample_id}_indel
      done

    pindel2vcf -r ${ref_dir}/${fasta} \
      -R GRCh38.d1.vd1 \
      -d GDC \
      -p pindel_result/${sample_id}_indel \
      -v pindel_result/${sample_id}.pindel.indel.vcf
    
  >>>
	
  runtime {
    docker: docker
    cluster: cluster
    systemDisk: "cloud_ssd 40"
    dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
  }

  output {
    Array[File] pindel_result = glob("./pindel_result/${sample_id}*")
  }
}