rnaseq_for_se/tasks/fastp.wdl

task fastp {
    File read
    String sample_id=sub(basename(read), "\\.(fastq|fq)\\.gz$", "")
    String adapter_sequence
    String adapter_sequence_r2
    String docker
    String cluster
    String disk_size
    String umi_loc	
    Int trim_front1
    Int trim_tail1
    Int max_len1
    Int trim_front2
    Int trim_tail2
    Int max_len2
    Int disable_adapter_trimming
    Int length_required
    Int umi_len
    Int UMI
    Int qualified_quality_phred
    Int length_required1
    Int disable_quality_filtering
   
	command <<<
        mkdir -p /cromwell_root/tmp/fastp/
	##1.Disable_quality_filtering
	if [ "${disable_quality_filtering}" == 0 ]
        then
	cp ${read} /cromwell_root/tmp/fastp/{sample_id}.fastq.tmp1.gz
        else
	fastp --thread 4 --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1} -i ${read} -o /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp1.gz -j ${sample_id}.json -h ${sample_id}.html
        fi

	##2.UMI
	if [ "${UMI}" == 0 ]
        then
	cp /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp1.gz /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp2.gz
        else
	fastp --thread 4 -U --umi_loc=${umi_loc} --umi_len=${umi_len} --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1} -i /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp1.gz -o /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp2.gz -j ${sample_id}.json -h ${sample_id}.html
	fi

	##3.Trim
        if [ "${disable_adapter_trimming}" == 0 ]
        then
	fastp --thread 4 -l ${length_required} -q ${qualified_quality_phred} -u ${length_required1} --adapter_sequence ${adapter_sequence}  --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1}  -i /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp2.gz -o ${sample_id}.fastq.gz -j ${sample_id}.json -h ${sample_id}.html
        else
	cp /cromwell_root/tmp/fastp/${sample_id}.fastq.tmp2.gz ${sample_id}.fastq.gz
        fi
   >>>
   
   runtime { 
		docker: docker
		cluster: cluster
		systemDisk: "cloud_ssd 40"
		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
   }

   output {
      File json = "${sample_id}.json"
      File report = "${sample_id}.html"
      File Trim = "${sample_id}.fastq.gz"
   }
}