pipeline/wdl/call_mutation.wdl

task mutation_calling_umi {
    String name
    String output_dir
    String rmdup_bam
    String ref
    String bed
    command <<<

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

        #1条call
        # 这个情况是reads数目只有1，但是如果去掉了这个reads数导致数据量减少很多
        # -r 3 是指有3条这样样的reads支撑
        # -f 是指频率 以2条方式的call出来的变异频率可以比1条的方式更可信

        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar \
        -G ${ref} \
        -f 0.001 \
        -N ${name} \
        -b ${rmdup_bam} \
        -UN -Q 20 -m 3 -r 3 -th 10 -z 1 -c 1 -S 2 -E 3 -g 4 ${bed} \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/teststrandbias.R \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_valid.pl \
        -N ${name} -E -f 0.001 > ${output_dir}/mutation/${name}.1r.snp.indel.vcf

        #提取>=2条矫正的序列
        bam_fetch.py ${output_dir}/alignment/${name}.rmdup.bam ${output_dir}/alignment/${name}.2r.rmdup.bam
        samtools index ${output_dir}/alignment/${name}.2r.rmdup.bam

        # 保证 1r call mut umi family 里面有2条reads
        #2条矫正的call
        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar -G ${ref} \
        -f 0.0001 -N ${name}_2r -b ${output_dir}/alignment/${name}.2r.rmdup.bam \
        -UN -Q 20 -m 3 -r 1 -th 10 -c 1 -S 2 -E 3 -g 4 ${bed} | /dataseq/jmdna/software/VarDict-1.8.3/bin/teststrandbias.R \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_valid.pl -N ${name} -E -f 0.001 >${output_dir}/mutation/${name}.2r.snp.indel.vcf

        #merge突变，以1条方式call的>0.01的突变+两条方式的对一条方式的低频区域（AF<0.01）进行矫正。
        correct_umi_1r_plus_2r.pl \
        ${output_dir}/mutation/${name}.1r.snp.indel.vcf \
        ${output_dir}/mutation/${name}.2r.snp.indel.vcf \
        ${output_dir}/mutation/${name}.snp.indel.raw.vcf

        correct_genome_3rule.py ${output_dir}/mutation/${name}.snp.indel.raw.vcf ${output_dir}/mutation/${name}.snp.indel.vcf ${ref}
        cp ${output_dir}/mutation/${name}.snp.indel.vcf ${output_dir}/mutation/${name}.snp.indel.somatic.vcf

        python ~/project/pipeline/workflow/script/tools/vcf_filter.py \
        -i ${output_dir}/mutation/${name}.snp.indel.somatic.vcf \
        -e 'INFO/AF[0] > 0.1' \
        -o ${output_dir}/mutation/${name}.snp.indel.germline.vcf

    >>>

    output {
        String somatic_vcf = "${output_dir}/mutation/${name}.snp.indel.somatic.vcf"
        String germline_vcf = "${output_dir}/mutation/${name}.snp.indel.germline.vcf"
    }
}

task mutation_calling_tissue {
    String name
    String bed
    String ref
    String output_dir
    String rmdup_bam

    command <<<
        if [ ! -d ${output_dir}/mutation ];then
        mkdir ${output_dir}/mutation
        fi

        # vardict
        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar \
        -G ${ref} \
        -f 0.01 \
        -N ${name} \
        -b ${rmdup_bam} \
        -UN \
        -Q 20 \
        -m 3 \
        -r 3 \
        -z 1 \
        -th 10 \
        -c 1 -S 2 -E 3 -g 4 ${bed} \
        |/dataseq/jmdna/software/VarDict-1.8.3/bin/teststrandbias.R \
        |/dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_valid.pl -N ${name} -E -f 0.01 \
        >${output_dir}/mutation/${name}.snp.indel.raw.vcf

        correct_genome_3rule.py ${output_dir}/mutation/${name}.snp.indel.raw.vcf ${output_dir}/mutation/${name}.snp.indel.vcf ${ref}
        cp ${output_dir}/mutation/${name}.snp.indel.vcf ${output_dir}/mutation/${name}.snp.indel.somatic.vcf

        python ~/project/pipeline/workflow/script/tools/vcf_filter.py \
        -i ${output_dir}/mutation/${name}.snp.indel.somatic.vcf \
        -e 'INFO/AF[0] > 0.1' \
        -o ${output_dir}/mutation/${name}.snp.indel.germline.vcf

    >>>

    output {
        String somatic_vcf = "${output_dir}/mutation/${name}.snp.indel.somatic.vcf"
        String germline_vcf = "${output_dir}/mutation/${name}.snp.indel.germline_vcf.vcf"
    }
}

task mutation_calling_tissue_control {
    String name
    String bed
    String ref
    String output_dir
    String tumor_rmdup_bam
    String normal_rmdup_bam

    command <<<
        if [ ! -d ${output_dir}/mutation ];then
        mkdir ${output_dir}/mutation
        fi

        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar \
        -G ${ref} \
        -f 0.01 \
        -N ${name} \
        -b "${tumor_rmdup_bam}|${normal_rmdup_bam}" \
        -UN \
        -Q 20 \
        -m 3 \
        -r 3 \
        -th 20 \
        -z 1 -c 1 -S 2 -E 3 -g 4 ${bed} | /dataseq/jmdna/software/VarDict-1.8.3/bin/testsomatic.R \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_paired.pl -N ${name} -f 0.01 \
        >${output_dir}/mutation/${name}.snp.indel.raw.vcf

        correct_genome_3rule.py ${output_dir}/mutation/${name}.snp.indel.raw.vcf ${output_dir}/mutation/${name}.snp.indel.vcf ${ref}

        python ~/project/pipeline/workflow/script/tools/vcf_filter.py -i ${output_dir}/mutation/${name}.snp.indel.vcf \
        -o ${output_dir}/mutation/${name}.snp.indel.germline.vcf \
        -e 'INFO/STATUS="Germline"'

        python ~/project/pipeline/workflow/script/tools/vcf_filter.py -i ${output_dir}/mutation/${name}.snp.indel.vcf \
        -o ${output_dir}/mutation/${name}.snp.indel.somatic.vcf \
        -e 'INFO/STATUS="StrongSomatic" | ( INFO/STATUS="LikelySomatic" && FORMAT/AF[0] > 3*FORMAT/AF[1] )'

    >>>

    output {
        String somatic_vcf = "${output_dir}/mutation/${name}.snp.indel.somatic.vcf"
        String germline_vcf = "${output_dir}/mutation/${name}.snp.indel.germline.vcf"
    }
}

task mutation_calling_umi_control {
    String name
    String bed
    String ref
    String output_dir
    String tumor_rmdup_bam
    String normal_rmdup_bam

    command <<<
        if [ ! -d ${output_dir}/mutation ];then
        mkdir ${output_dir}/mutation
        fi

        # 对照样本
        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar \
        -G ${ref} \
        -f 0.01 \
        -N ${name} \
        -b ${normal_rmdup_bam} \
        -UN \
        -Q 20 \
        -m 3 \
        -r 3 \
        -th 10 \
        -c 1 -S 2 -E 3 -g 4 ${bed} |/dataseq/jmdna/software/VarDict-1.8.3/bin/teststrandbias.R \
        |/dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_valid.pl -N ${name} -E -f 0.01 >${output_dir}/mutation/${name}.snp.indel.raw_germline.vcf

        # 实验样本
        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar \
        -G ${ref} \
        -f 0.001 \
        -N ${name} \
        -b ${tumor_rmdup_bam} \
        -UN -Q 20 -m 3 -r 3 -th 10 -c 1 -S 2 -E 3 -g 4 ${bed} \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/teststrandbias.R \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_valid.pl \
        -N ${name} -E -f 0.001 > ${output_dir}/mutation/${name}.1r.snp.indel.vcf

        #提取>=2条矫正的序列
        func_fetch_bam.py ${output_dir}/alignment/${name}.rmdup.bam ${output_dir}/alignment/${name}.2r.rmdup.bam
        samtools index ${output_dir}/alignment/${name}.2r.rmdup.bam

        # 保证 1r call mut umi family 里面有2条reads
        #2条矫正的call
        java -jar /dataseq/jmdna/software/VarDict-1.8.3/lib/VarDict-1.8.3.jar -G ${ref} \
        -f 0.0001 -N ${name}_2r -b ${output_dir}/alignment/${name}.2r.rmdup.bam \
        -UN -Q 20 -m 3 -r 1 -th 10 -c 1 -S 2 -E 3 -g 4 ${bed} | /dataseq/jmdna/software/VarDict-1.8.3/bin/teststrandbias.R \
        | /dataseq/jmdna/software/VarDict-1.8.3/bin/var2vcf_valid.pl -N ${name} -E -f 0.001 >${output_dir}/mutation/${name}.2r.snp.indel.vcf

        # merge突变，以1条方式call的>0.01的突变+两条方式的对一条方式的低频区域（AF<0.01）进行矫正。
        filter_snpindel_umi_1r_plus_2r.pl \
        ${output_dir}/mutation/${name}.1r.snp.indel.vcf \
        ${output_dir}/mutation/${name}.2r.snp.indel.vcf \
        ${output_dir}/mutation/${name}.snp.indel.pre_raw.vcf

        # 去除normal 中的突变位点
        correct_umi_subnormal.pl \
        ${output_dir}/mutation/${name}.snp.indel.pre_raw.vcf \
        ${output_dir}/mutation/${name}.snp.indel.raw_germline.vcf \
        ${output_dir}/mutation/${name}.snp.indel.raw_somaitc.vcf

        correct_genome_3rule.py ${output_dir}/mutation/${name}.snp.indel.raw_germline.vcf ${output_dir}/mutation/${name}.snp.indel.germline.vcf ${ref}
        correct_genome_3rule.py ${output_dir}/mutation/${name}.snp.indel.raw_somaitc.vcf ${output_dir}/mutation/${name}.snp.indel.somatic.vcf ${ref}

    >>>

    output {
        String somatic_vcf = "${output_dir}/mutation/${name}.snp.indel.somatic.vcf"
        String germline_vcf = "${output_dir}/mutation/${name}.snp.indel.germline.vcf"
    }
}

task annovar {
    String prefix
    String output_dir
    String ref
    String vcf

    command <<<

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

        table_annovar.pl \
        ${vcf} \
        /dataseq/jmdna/software/annovar/humandb/ \
        -buildver hg19 -nastring . -vcfinput -remove -otherinfo \
        -protocol refGene,avsnp150,cosmic91,clinvar_20220320,1000g2015aug_all,1000g2015aug_eas,esp6500siv2_all,exac03nontcga,gnomad_genome,dbnsfp35c,cytoBand \
        -argument '-splicing_threshold 2 -hgvs',,,,,,,,,, \
        -intronhgvs 50 \
        -operation g,f,f,f,f,f,f,f,f,f,r \
        -outfile ${output_dir}/mutation/${prefix} \
        -dot2underline

    >>>
    output {
        String anno = "${output_dir}/mutation/${prefix}.hg19_multianno.txt"
    }
}

task filter_umi {
    String name
    String anno
    String project
    String output_dir
    String tumor_rmdup_bam

    command <<<

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

        filter_snpindel \
        ${anno} \
        ${project} \
        c \
        ${output_dir}/mutation/${name}.snp.indel.somatic.hg19_multianno.filtered.pre.txt \
        ${output_dir}/mutation/${name}.snp.indel.germline.hg19_multianno.filtered.pre.txt \
        ${output_dir}/mutation/${name}.snp.indel.hg19_multianno.tag.txt

        filter_snpindel_umi_correct_overlap_reads.py \
        ${output_dir}/mutation/${name}.snp.indel.somatic.hg19_multianno_filtered_pre.txt \
        ${tumor_rmdup_bam} \
        ${output_dir}/mutation/${name}.snp.indel.somatic.hg19_multianno_filtered.txt

    >>>
    output {
        String snvindel_filtered= "${output_dir}/mutation/${name}.snp.indel.Somatic.anno.hg19_multianno_filtered.txt"
        String germline_filtered = "${output_dir}/mutation/${name}.snp.indel.Germline.anno.hg19_multianno_filtered.txt"
    }
}

task filter_tissue {
    String name
    String anno
    String project
    String output_dir
    String tumor_rmdup_bam

    command <<<

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

        filter_snpindel.pl \
        ${anno} \
        ${project} \
        t \
        ${output_dir}/mutation/${name}.snp.indel.somatic.hg19_multianno.filtered.txt \
        ${output_dir}/mutation/${name}.snp.indel.germline.hg19_multianno.filtered.txt \
        ${output_dir}/mutation/${name}.snp.indel.hg19_multianno.tag.txt

    >>>
    output {
        String snvindel_filtered= "${output_dir}/mutation/${name}.snp.indel.Somatic.anno.hg19_multianno_filtered.txt"
        String germline_filtered = "${output_dir}/mutation/${name}.snp.indel.Germline.anno.hg19_multianno_filtered.txt"
    }
}

workflow call_mutation {

    String tumor
    String tumor_rmdup_bam
    String? normal
    String? normal_rmdup_bam
    Boolean umi
    String output_dir
    String ref
    String bed

    # pipe 执行 mutation_calling => annovar => filter

    # 单样本
    if (!defined(normal)) {
        if (umi) {
            call mutation_calling_umi {
                input:
                    name=tumor,
                    output_dir=output_dir,
                    ref=ref,
                    bed=bed,
                    rmdup_bam=tumor_rmdup_bam
            }
            call annovar as anno_somatic_umi {
                input:
                    prefix="${tumor}.snp.indel.somatic",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_umi.somatic_vcf
            }
            call annovar as anno_germline_umi {
                input:
                    prefix="${tumor}.snp.indel.germline",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_umi.germline_vcf
            }
        }

        if (!umi) {
            call mutation_calling_tissue {
                input:
                    name=tumor,
                    output_dir=output_dir,
                    ref=ref,
                    bed=bed,
                    rmdup_bam=tumor_rmdup_bam
            }
            call annovar as anno_somatic_tissue {
                input:
                    prefix="${tumor}.snp.indel.somatic",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_tissue.somatic_vcf
            }
            call annovar as anno_germline_tissue {
                input:
                    prefix="${tumor}.snp.indel.germline",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_tissue.germline_vcf
            }
        }
    }
    # 双样本
    if (defined(normal)) {
        if (umi) {
            call mutation_calling_umi_control {
                input:
                    name=tumor,
                    output_dir=output_dir,
                    ref=ref,
                    bed=bed,
                    tumor_rmdup_bam=tumor_rmdup_bam,
                    normal_rmdup_bam=normal_rmdup_bam

            }
            call annovar as anno_somatic_umi_control {
                input:
                    prefix="${tumor}.snp.indel.somatic",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_umi_control.somatic_vcf
            }
            call annovar as anno_germline_umi_control {
                input:
                    prefix="${tumor}.snp.indel.germline",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_umi_control.germline_vcf
            }
        }
        if (!umi) {
            call mutation_calling_tissue_control {
                input:
                    name=tumor,
                    output_dir=output_dir,
                    ref=ref,
                    bed=bed,
                    tumor_rmdup_bam=tumor_rmdup_bam,
                    normal_rmdup_bam=normal_rmdup_bam

            }
            call annovar as anno_somatic_tissue_control {
                input:
                    prefix="${tumor}.snp.indel.somatic",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_tissue_control.somatic_vcf
            }
            call annovar as anno_germline_tissue_control {
                input:
                    prefix="${tumor}.snp.indel.germline",
                    output_dir=output_dir,
                    ref=ref,
                    vcf=mutation_calling_tissue_control.germline_vcf
            }
        }
    }

    output {
        String somatic_vcf = "${output_dir}/mutation/${tumor}.snp.indel.somatic.vcf"
        String germline_vcf = "${output_dir}/mutation/${tumor}.snp.indel.germline.vcf"
    }
}