pipeline/wdl/neoantigen.wdl

task run_neoantigen {
    String tumor
    String? normal
    String input_dir
    String output_dir
    String ref
    String tumor_rmdup_bam
    String somatic_vcf
    String germline_vcf
    String sample_type

    command <<<

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

        razers3 -tc 10 -i 95 -m 1 -dr 0 -o ${output_dir}/neoantigen/hla/fished_1.bam \
        /dataseq/jmdna/software/OptiType-1.3.5/data/hla_reference_dna.fasta  ${input_dir}/*_${normal}_*1.fq.gz
        samtools bam2fq ${output_dir}/neoantigen/hla/fished_1.bam > ${output_dir}/neoantigen/hla/${normal}_1_fished.fastq
        rm ${output_dir}/neoantigen/hla/fished_1.bam
        razers3 -tc 10 -i 95 -m 1 -dr 0 -o ${output_dir}/neoantigen/hla/fished_2.bam \
        /dataseq/jmdna/software/OptiType-1.3.5/data/hla_reference_dna.fasta ${input_dir}/*_${normal}_*2.fq.gz
        samtools bam2fq ${output_dir}/neoantigen/hla/fished_2.bam > ${output_dir}/neoantigen/hla/${normal}_2_fished.fastq
        rm ${output_dir}/neoantigen/hla/fished_2.bam
        /dataseq/jmdna/software/OptiType-1.3.5/OptiTypePipeline.py \
        -i ${output_dir}/neoantigen/hla/${normal}_1_fished.fastq  \
        ${output_dir}/neoantigen/hla/${normal}_2_fished.fastq --dna -v \
        --prefix ${normal} -o ${output_dir}/neoantigen/hla/

        #step1:vep annotation and variant filter
        vep \
        --input_file ${somatic_vcf} \
        --output_file ${output_dir}/neoantigen/${tumor}_somatic_vepanno.vcf \
        --fasta /home/install/ref/hg19/ \
        --dir /dataseq/jmdna/software/.vep/ \
        --format vcf \
        --vcf \
        --symbol \
        --terms SO \
        --tsl \
        --hgvs \
        --offline \
        --cache \
        --plugin Downstream \
        --plugin Wildtype \
        --plugin Frameshift \
        --pick \
        --transcript_version 100 \
        --force_overwrite

        vep \
        --input_file ${germline_vcf} \
        --output_file  ${output_dir}/neoantigen/${tumor}_germline_vepanno.vcf \
        --fasta /home/install/ref/hg19/ \
        --offline --cache --dir /dataseq/jmdna/software/.vep/ \
        --format vcf \
        --vcf \
        --symbol \
        --terms SO \
        --tsl \
        --hgvs \
        --plugin Downstream --plugin Wildtype --plugin Frameshift \
        --pick \
        --transcript_version 100 \
        --force_overwrite

        filter_vep \
        -i ${output_dir}/neoantigen/${tumor}_somatic_vepanno.vcf \
        -o ${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter.vcf \
        --format vcf \
        --force_overwrite \
        --filter "(Consequence != synonymous_variant) \
        and (Consequence != intron_variant) \
        and (Consequence != intergenic_variant) \
        and (Consequence != 3_prime_UTR_variant) \
        and (Consequence != upstream_gene_variant) \
        and (Consequence != downstream_gene_variant) \
        and (Consequence != 5_prime_UTR_variant)"

        filter_vep \
        -i ${output_dir}/neoantigen/${tumor}_germline_vepanno.vcf \
        -o ${output_dir}/neoantigen/${tumor}_germline_vepanno_vepfilter.vcf \
        --format vcf \
        --force_overwrite \
        --filter "(Consequence != synonymous_variant) \
        and (Consequence != intron_variant) \
        and (Consequence != intergenic_variant) \
        and (Consequence != 3_prime_UTR_variant) \
        and (Consequence != upstream_gene_variant) \
        and (Consequence != downstream_gene_variant) \
        and (Consequence != 5_prime_UTR_variant)"

        filter_neoantigen.pl somatic \
        ${tumor} \
        ${sample_type} \
        ${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter.vcf \
        ${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter_filter.vcf

        filter_neoantigen.pl germline \
        ${tumor} \
        ${sample_type} \
        ${output_dir}/neoantigen/${tumor}_germline_vepanno_vepfilter.vcf \
        ${output_dir}/neoantigen/${tumor}_germline_vepanno_vepfilter_filter.vcf

        #step2 phasing variant

        java -Xmx16g -jar $PICARD SortVcf \
        I=${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter_filter.vcf \
        I=${output_dir}/neoantigen/${tumor}_germline_vepanno_vepfilter_filter.vcf \
        O=${output_dir}/neoantigen/${tumor}_combined.sorted.vcf \
        SEQUENCE_DICTIONARY=/home/install/ref/hg19/hg19.dict

        java -Xmx16g -jar /dataseq/jmdna/software/GenomeAnalysisTK.3.7.jar \
        -T ReadBackedPhasing \
        -R ${ref} \
        -I ${tumor_rmdup_bam} \
        --variant ${output_dir}/neoantigen/${tumor}_combined.sorted.vcf \
        -L ${output_dir}/neoantigen/${tumor}_combined.sorted.vcf \
        --cacheWindowSize 90 \
        -o ${output_dir}/neoantigen/${tumor}.phased.vcf

        ##step3:bgzip and index the VCF
        bgzip -c ${output_dir}/neoantigen/${tumor}.phased.vcf > ${output_dir}/neoantigen/${tumor}.phased.vcf.gz
        tabix -p vcf ${output_dir}/neoantigen/${tumor}.phased.vcf.gz

        bgzip -c ${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter_filter.vcf > ${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter_filter.vcf.gz
        tabix -p vcf ${output_dir}/neoantigen/${tumor}_somatic_vepanno_vepfilter_filter.vcf.gz

        HLA=`head -n 2 ${output_dir}/neoantigen/hla/*tsv | tail -n 1 | awk '{OFS=","}{print "HLA-"$2,"HLA-"$3,"HLA-"$4,"HLA-"$5,"HLA-"$6,"HLA-"$7}'|xargs echo -n`

        #IEDB recommed NetMHCpan. but NetMHCcons is also outperformed(PMID: 31204427)
        #8,9,10,11
        docker run --rm -u $UID:$(id -g $UID) -v ${output_dir}/neoantigen/:/data griffithlab/pvactools:2.0.7 pvacseq run \
        -e1 9,10,11 \
        --iedb-install-directory /opt/iedb/ \
        -t 10 \
        -k \
        --normal-sample-name NORMAL \
        -p /data/${tumor}.phased.vcf.gz \
        /data/${tumor}_somatic_vepanno_vepfilter_filter.vcf.gz \
        ${tumor} \
        $HLA \
        NetMHCpan \
        /data

        dos2unix ${output_dir}/neoantigen/MHC_Class_I/*.all_epitopes.tsv
        #        perl ${output_dir}/netchop.pl $outputDir $name $tumor $max_peptide_length
    >>>

}

workflow call_neoantigen {

    Boolean run=true

    String tumor
    String? normal
    String input_dir
    String output_dir
    String ref
    String tumor_rmdup_bam
    String somatic_vcf
    String germline_vcf
    Boolean umi

    if (run) {
        call run_neoantigen {
            input:
                tumor=tumor,
                normal=normal,
                input_dir=input_dir,
                output_dir=output_dir,
                ref=ref,
                tumor_rmdup_bam=tumor_rmdup_bam,
                somatic_vcf=somatic_vcf,
                germline_vcf=germline_vcf,
                sample_type=if umi then 'c' else 't'
        }
    }
}