gatk4/tasks/gatk.wdl

task gatk {
    String sample_id
    File normal_bam
    File normal_bai
    File gatk4_database

    String docker
    String cluster
    String disk_size

    command <<<

        set -o pipefail
        set -e
        echo "gatk start!"
        cp -r ${gatk4_database} ./gatk4_database
        cp ${normal_bam} ./${sample_id}.bam
        cp ${normal_bai} ./${sample_id}.bam.bai
        hg38_genome="./gatk4_database/GRCh38.d1.vd1.fa"
        dbsnp="./gatk4_database/resources_broad_hg38_v0_Homo_sapiens_assembly38.dbsnp138.vcf.gz"
        indel1="./gatk4_database/resources_broad_hg38_v0_Homo_sapiens_assembly38.known_indels.vcf.gz"
        indel2="./gatk4_database/resources_broad_hg38_v0_Mills_and_1000G_gold_standard.indels.hg38.vcf.gz"
        bed="./gatk4_database/aim_bed_use_sort.bed"
        pon="./gatk4_database/somatic-hg38_1000g_pon.hg38.vcf.gz"
        EXAC="./gatk4_database/somatic-hg38_small_exac_common_3.hg38.vcf.gz"
        
        mkdir markduplicate
        mkdir BQSR
        mkdir Mutect2
        mkdir gatk4_resource

        java -jar -Xmx40g /opt/picard.jar MarkDuplicates INPUT=./${sample_id}.bam OUTPUT=./markduplicate/${sample_id}_marked.bam METRICS_FILE=./markduplicate/${sample_id}.metrics
        echo "dedup done"

        samtools index -@ 6 ./markduplicate/${sample_id}_marked.bam
        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" BaseRecalibrator -I ./markduplicate/${sample_id}_marked.bam  -R $hg38_genome  -O ./BQSR/${sample_id}_markdup_bqsr.report   --known-sites $dbsnp   --known-sites $indel1   --known-sites $indel2 -L $bed

        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" ApplyBQSR  -I ./markduplicate/${sample_id}_marked.bam  -R $hg38_genome -L $bed --bqsr-recal-file ./BQSR/${sample_id}_markdup_bqsr.report   -O ./BQSR/${sample_id}_markdup_bqsr.bam
        echo "BQSR done"

        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" Mutect2 -R $hg38_genome -I ./BQSR/${sample_id}_markdup_bqsr.bam  -L $bed --pon $pon --germline-resource $dbsnp --tumor-lod-to-emit 2 --f1r2-tar-gz Mutect2/${sample_id}.f1r2.tar.gz  -O Mutect2/${sample_id}_mutect2_raw.vcf
        echo "Mutect2 done"

        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" GetPileupSummaries -R $hg38_genome -I ./BQSR/${sample_id}_markdup_bqsr.bam  -V $EXAC -L $EXAC -L $bed --interval-set-rule INTERSECTION  -O Mutect2/${sample_id}.getpileupsum.table

        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" CalculateContamination -I Mutect2/${sample_id}.getpileupsum.table  -O Mutect2/${sample_id}.contamination.table --tumor-segmentation Mutect2/${sample_id}.segmentation.table

        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" LearnReadOrientationModel -I Mutect2/${sample_id}.f1r2.tar.gz  -O Mutect2/${sample_id}_f1r2_artifact.tar.gz

        gatk --java-options "-Djava.io.tmpdir=./BQSR -Xms40G -Xmx40G -XX:ParallelGCThreads=2" FilterMutectCalls  -V Mutect2/${sample_id}_mutect2_raw.vcf  -R $hg38_genome  --contamination-table Mutect2/${sample_id}.contamination.table --tumor-segmentation Mutect2/${sample_id}.segmentation.table  --ob-priors Mutect2/${sample_id}_f1r2_artifact.tar.gz  -O Mutect2/${sample_id}_mutect2_filtered.vcf
        bcftools norm -f $hg38_genome -O v  -o Mutect2/${sample_id}_mutect2_filtered_bcfed.vcf  Mutect2/${sample_id}_mutect2_filtered.vcf


    >>>

    runtime {
        docker : docker
        cluster: cluster
        systemDisk: "cloud_ssd 100"
        dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
    }

    output {
        File raw_vcf = "./Mutect2/${sample_id}_mutect2_raw.vcf"
        File filtered_vcf = "./Mutect2/${sample_id}_mutect2_filtered.vcf"
    }

}