pipeline/wdl/qc.wdl

task runqc {
    String name
    String input_dir
    String output_dir

    command <<<
        # echo "###### ${name} fastp beginning at: $(date) ######"

        if [ ! -d ${output_dir}/qc ];then
        mkdir ${output_dir}/qc
        fi

        fastp \
        -i ${input_dir}/*_${name}_*1.fq.gz \
        -o ${output_dir}/qc/${name}_clean_R1.fq.gz \
        -I ${input_dir}/*_${name}_*2.fq.gz \
        -O ${output_dir}/qc/${name}_clean_R2.fq.gz \
        -w 10 \
        -e 20 \
        --correction \
        --overlap_len_require 10 \
        -j ${output_dir}/qc/${name}.json \
        -h ${output_dir}/qc/${name}.html \
        --report_title ${name}

    >>>
    output {
        String r1 = "${output_dir}/qc/${name}_clean_R1.fq.gz "
        String r2 = "${output_dir}/qc/${name}_clean_R2.fq.gz "
        String json = "${output_dir}/qc/${name}.json"
    }

}

task umiqc {
    String name
    String input_dir
    String output_dir

    command <<<
        # echo "###### ${name} fastp beginning at: $(date) ######"

        if [ ! -d ${output_dir}/qc ];then
        mkdir ${output_dir}/qc
        fi

        fastp -i ${input_dir}/*_${name}_*1.fq.gz -o ${output_dir}/qc/${name}_clean_R1.fq.gz \
        -I ${input_dir}/*_${name}_*2.fq.gz -O ${output_dir}/qc/${name}_clean_R2.fq.gz \
        -w 10 \
        --correction \
        --overlap_len_require=10 \
        --umi \
        --umi_loc=per_read \
        --umi_len=4 \
        --umi_prefix=UMI \
        --umi_skip=3 \
        --umi_delim \ : \
        --disable_trim_poly_g \
        -j ${output_dir}/qc/${name}.json \
        -h ${output_dir}/qc/${name}.html \
        --report_title ${name}

    >>>

    output {
        String r1 = "${output_dir}/qc/${name}_clean_R1.fq.gz "
        String r2 = "${output_dir}/qc/${name}_clean_R2.fq.gz "
        String json = "${output_dir}/qc/${name}.json"
    }

}

workflow qc {

    String tumor
    String? normal
    Boolean umi
    String input_dir
    String output_dir

    scatter(name in [tumor, normal]) {
        if (defined(name)) {
            if (name==tumor) {
                if (umi) {
                    call umiqc as run_umi_qc {
                        input:
                            name=name,
                            input_dir=input_dir,
                            output_dir=output_dir
                    }
                }
                if (!umi) {
                    call runqc as run_tumor_qc {
                        input:
                            name=name,
                            input_dir=input_dir,
                            output_dir=output_dir
                    }
                }
            }
            if (name==select_first([normal, 'None'])) {
                call runqc as run_normal_qc {
                    input:
                        name=name,
                        input_dir=input_dir,
                        output_dir=output_dir
                }
            }

        }
    }

    output {
        String tumor_r1 = "${output_dir}/qc/${tumor}_clean_R1.fq.gz "
        String tumor_r2 = "${output_dir}/qc/${tumor}_clean_R2.fq.gz "
        String tumor_json = "${output_dir}/qc/${tumor}.json"
        String normal_r1 = "${output_dir}/qc/${normal}_clean_R1.fq.gz "
        String normal_r2 = "${output_dir}/qc/${normal}_clean_R2.fq.gz "
        String normal_json = "${output_dir}/qc/${normal}.json"
    }
}