pipeline/wdl/fusion.wdl

task rmdup_picard {
    String name
    String sorted_bam
    String output_dir
    String ref

    command <<<

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

        java -XX:+UseParallelGC -XX:ParallelGCThreads=8 -Xmx12G  -jar $PICARD MarkDuplicates \
        I=${sorted_bam} \
        O=${output_dir}/alignment/${name}.picard.rmdup.bam  \
        CREATE_INDEX=true \
        M=${output_dir}/alignment/${name}.picard.rmdup.metrics.txt  \
        R=${ref}

    >>>
    output {
        String rmdup_bam = "${output_dir}/alignment/${name}.picard.rmdup.bam"
    }

}

task fusion_calling {
    String name
    String rmdup_bam
    String output_dir
    String ref
    String tumor_bamdst_depth

    command <<<

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

        # Extract the discordant paired-end alignments.
        samtools view -b -F 1294 ${rmdup_bam} > ${output_dir}/fusion/${name}.discordants.bam

        # Extract the split-read alignments
        samtools view -h ${rmdup_bam} \
        | /dataseq/jmdna/software/lumpy-sv/scripts/extractSplitReads_BwaMem -i stdin \
        | samtools view -Sb - \
        > ${output_dir}/fusion/${name}.splitters.bam

        lumpyexpress \
        -B ${rmdup_bam} \
        -S ${output_dir}/fusion/${name}.splitters.bam \
        -D ${output_dir}/fusion/${name}.discordants.bam \
        -o ${output_dir}/fusion/${name}.fusion.raw.vcf

        vcf_filter.py \
        -i ${output_dir}/fusion/${name}.fusion.raw.vcf \
        -e 'INFO/PE[0] >1 && INFO/SR[0] > 4 ' \
        | svtyper -B ${rmdup_bam} -T ${ref} -o ${output_dir}/fusion/${name}.fusion.vcf

        table_annovar.pl \
        ${output_dir}/fusion/${name}.fusion.vcf \
        /dataseq/jmdna/software/annovar/humandb/ \
        -buildver hg19 -nastring . -vcfinput -remove -otherinfo \
        -protocol refGene \
        -operation g \
        -outfile ${output_dir}/fusion/${name}.fusion

        filter_fusion.pl ${tumor_bamdst_depth} \
        ${output_dir}/fusion/${name}.fusion.hg19_multianno.vcf \
        ${output_dir}/fusion/${name}.fusion.hg19_multianno.filter.txt

        vcf_filter.py \
        -e 'INFO/SVTYPE = "DEL" | INFO/SVTYPE = "DUP" | INFO/SVTYPE = "INS" ' \
        -i ${output_dir}/fusion/${name}.fusion.hg19_multianno.vcf \
        -o ${output_dir}/fusion/${name}.longindel.hg19_multianno.vcf \

    >>>

    output {
        String vcf_txt = "${output_dir}/fusion/${name}.fusion.hg19_multianno.filter.txt"
        String longvcf = "${output_dir}/fusion/${name}.longindel.hg19_multianno.vcf"
    }

}

task fusion_sum {
    String name
    String output_dir
    String vcf_txt
    String longvcf
    String project
    String cancer

    command <<<
        target_therapy_fusion.pl \
        ${vcf_txt} \
        ${longvcf} \
        ${output_dir}/fusion/${name}.fusion.hg19_multianno.filter \
        ${output_dir}/fusion/${name}.longindel.hg19_multianno.longindel.pos.txt ${project}  ${cancer}
    >>>
    output {
        String fusion = "${output_dir}/fusion/${name}.fusion.hg19_multianno.filter.txt"
    }
}

workflow call_fusion {

    String name
    String sorted_bam
    String rmdup_bam
    String output_dir
    String ref
    Boolean umi
    String tumor_bamdst_depth
    String project
    String cancer

    if (umi) {
        call rmdup_picard {
            input:
                name=name,
                sorted_bam=sorted_bam,
                output_dir=output_dir,
                ref=ref
        }
        call fusion_calling as fusion_calling_umi {
            input:
                name=name,
                rmdup_bam=rmdup_picard.rmdup_bam,
                output_dir=output_dir,
                ref=ref,
                tumor_bamdst_depth=tumor_bamdst_depth
        }

    }

    if (!umi) {
        call fusion_calling as fusion_calling {
            input:
                name=name,
                rmdup_bam=rmdup_bam,
                output_dir=output_dir,
                ref=ref,
                tumor_bamdst_depth=tumor_bamdst_depth
        }
        call fusion_sum as fusion_sum {
            input:
                vcf_txt=fusion_calling.vcf_txt,
                longvcf=fusion_calling.longvcf,
                name=name,
                output_dir=output_dir,
                cancer=cancer,
                project=project

        }
    }

    output {
        String fusion_vcf = "${output_dir}/fusion/${name}.fusion.vcf"
    }
}