上传文件至 'tasks'

tasks/TNscope.wdl

@@ -0,0 +1,53 @@
+task sentieon_TNscope{
+    String sample_id
+    File tumor_bam
+    File tumor_bam_bai
+    File? normal_bam
+    File? normal_bam_bai
+    String tumor_name
+    String normal_name
+    File tumor_recall_data
+    File normal_recall_data
+
+    File ref_dir
+    String ref_fasta
+    File dbsnp_dir
+    String dbsnp
+
+    # excute env
+    String docker
+    String cluster_config
+    String disk_size
+    String SENTIEON_LICENSE
+
+
+    command <<<
+        set -o pipefail
+        set -exo
+        export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+        nt=$(nproc)
+
+        sentieon driver -t $nt -r ${ref_dir}/${ref_fasta} \
+        -i ${tumor_bam} -q ${tumor_recall_data} \
+        -i ${normal_bam} -q ${normal_recall_data} \
+        --algo TNscope --tumor_sample ${tumor_name} --normal_sample ${normal_name} \
+        --disable_detector sv --trim_soft_clip \
+        --dbsnp ${dbsnp_dir}/${dbsnp} ${sample_id}.TNscope.vcf || { echo "TNscope failed"; exit 1; }
+        
+        ls ./
+
+    >>>
+
+    runtime{
+        docker:docker
+        cluster:cluster_config
+        systemDisk:"cloud_ssd 40"
+        dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
+    }
+
+    output{
+        File vcf = "${sample_id}.TNscope.vcf"
+        File vcf_index = "${sample_id}.TNscope.vcf.idx"
+
+    }
+}

tasks/TNseq.wdl

@@ -0,0 +1,76 @@
+task sentieon_TNseq{
+    String sample_id
+    File tumor_bam
+    File tumor_bam_bai
+    File? normal_bam
+    File? normal_bam_bai
+    String tumor_name
+    String normal_name
+
+    File ref_dir
+    String ref_fasta
+    File germline_resource
+    File germline_resource_tbi
+
+    # excute env
+    String docker
+    String cluster_config
+    String disk_size
+    String SENTIEON_LICENSE
+
+
+    command <<<
+        set -o pipefail
+        set -exo
+        export SENTIEON_LICENSE=${SENTIEON_LICENSE}
+        nt=$(nproc)
+
+
+
+        INPUT="-i ${tumor_bam} -i ${normal_bam}"
+        SAMPLE="--tumor_sample ${tumor_name} --normal_sample ${normal_name}"
+
+
+        sentieon driver -t $nt -r ${ref_dir}/${ref_fasta} \
+        $INPUT \
+        --algo TNhaplotyper2 $SAMPLE \
+        --germline_vcf ${germline_resource} \
+        ${sample_id}.TNseq.raw.vcf \
+        --algo OrientationBias --tumor_sample ${tumor_name} \
+        ${sample_id}.orientation \
+        --algo ContaminationModel $SAMPLE \
+        --vcf ${germline_resource} \
+        --tumor_segments ${sample_id}.contamination.segments \
+        ${sample_id}.contamination
+
+        sentieon driver -t $nt \
+        -r ${ref_dir}/${ref_fasta} \
+        --algo TNfilter $SAMPLE \
+        -v ${sample_id}.TNseq.raw.vcf \
+        --contamination ${sample_id}.contamination \
+        --tumor_segments ${sample_id}.contamination.segments \
+        --orientation_priors ${sample_id}.orientation \
+        ${sample_id}.bwa_TNseq.vcf
+
+    >>>
+
+    runtime{
+        docker:docker
+        cluster:cluster_config
+        systemDisk:"cloud_ssd 40"
+        dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
+
+    }
+
+    output{
+        File raw_vcf = "${sample_id}.TNseq.raw.vcf"
+        File raw_vcf_index = "${sample_id}.TNseq.raw.vcf.idx"
+        File vcf = "${sample_id}.bwa_TNseq.vcf"
+        File vcf_index = "${sample_id}.bwa_TNseq.vcf.idx"
+        File contamination = "${sample_id}.contamination"
+        File contamination_segments = "${sample_id}.contamination.segments"
+        File orientation = "${sample_id}.orientation"
+
+
+    }
+}

tasks/strelka.wdl

@@ -0,0 +1,51 @@
+task strelka_calling{
+    File tumor_bam
+    File tumor_bam_bai
+    File normal_bam
+    File normal_bam_bai
+    String ref_fasta
+    File ref_dir
+    String sample_id
+    File? manta_indel_vcf
+    File? manta_indel_vcf_index
+    
+    # excute env
+    String docker
+    String cluster_config
+    String disk_size
+    
+    String out_dir = "${sample_id}_result"
+    command <<<
+    set -exo pipefail
+    nt=$(nproc)
+    /home/biosoft/strelka-2.9.10.centos6_x86_64/bin/configureStrelkaSomaticWorkflow.py \
+    --normalBam ${normal_bam} \
+    --tumorBam ${tumor_bam} \
+    --referenceFasta ${ref_dir}/${ref_fasta} \
+    --runDir ${out_dir}
+    
+    ls ${out_dir}
+
+    python2.7 ${out_dir}/runWorkflow.py -m local -j $nt
+
+    ls ${out_dir}
+
+    tar cvf ${out_dir}.tar ${out_dir}
+    >>>
+
+    runtime{
+		docker:docker
+		cluster:cluster_config
+		systemDisk:"cloud_ssd 40"
+		dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
+    }
+
+    output{
+        File out_file = "${out_dir}.tar"
+        File indel_vcf = "${out_dir}/results/variants/somatic.indels.vcf.gz"
+        File indel_vcf_index = "${out_dir}/results/variants/somatic.indels.vcf.gz.tbi"
+        File snv_vcf = "${out_dir}/results/variants/somatic.snvs.vcf.gz"
+        File snv_vcf_index = "${out_dir}/results/variants/somatic.snvs.vcf.gz.tbi"
+        
+    }
+}

tasks/tmb.wdl

@@ -0,0 +1,45 @@
+task TMB {
+  
+  String sample
+  File hg38_CDS
+  File regions
+  File snpindel_txt
+  String docker
+  String cluster_config
+  String disk_size
+  Boolean AF_filter
+  
+  command <<<
+    set -o pipefail
+    set -e
+    nt=$(nproc)
+    
+    if [ ${regions} ];then
+        sort -k1,1 -k2,2n ${regions} | bedtools merge -i - > merged.bed
+        bedtools intersect -a merged.bed -b ${hg38_CDS} > merged_intersect.bed
+        size=`awk -F'\t' 'BEGIN{SUM=0}{SUM+=$3-$2}END{print SUM}' merged_intersect.bed`
+    else
+        size=`awk -F'\t' 'BEGIN{SUM=0}{SUM+=$3-$2}END{print SUM}' ${hg38_CDS}`
+    fi 
+    
+    # analysis in python
+    if ${AF_filter};then
+        python ~/tmb.py ${snpindel_txt} $size ${sample} 'yes'
+    else
+        python ~/tmb.py ${snpindel_txt} $size ${sample} 'no'
+    fi
+
+  >>>
+  
+  runtime {
+    docker: docker
+    cluster: cluster_config
+    systemDisk: "cloud_ssd 40"
+    dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+  }
+  
+  output {
+    File snp_indel="${sample}.snp_indel.txt"
+    File tmb="${sample}.TMB.txt"
+  }
+}