|
- task benchmark {
- File vcf
- File benchmarking_dir
- File sdf
- String project
- String sample = basename(vcf,".splited.vcf")
- String docker
- String cluster_config
- String disk_size
-
- command <<<
- set -o pipefail
- set -e
- nt=$(nproc)
- mkdir -p /cromwell_root/tmp
- cp -r ${benchmarking_dir} /cromwell_root/tmp/
-
- cat ${vcf} | grep '#' > header
- cat ${vcf} | grep -v '#' | grep -v '0/0' | grep -v '\./\.'| awk '
- BEGIN { OFS = "\t" }
- {
- for ( i=9; i<=NF; i++ ) {
- split($i,a,":") ;$i = a[1];
- }
- }
- { print }
- ' > body
- cat header body > filtered.vcf
-
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip filtered.vcf -c > ${sample}.rtg.vcf.gz
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${sample}.rtg.vcf.gz
-
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${project}.${sample}.rtg.vcf.gz -o ${sample}.rtg.SNV.vcf.gz --snps-only
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${project}.${sample}.rtg.vcf.gz -o ${sample}.rtg.INDEL.vcf.gz --non-snps-only
-
- if [[ ${sample} =~ "LCL5" ]];then
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- elif [[ ${sample} =~ "LCL6" ]]; then
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- elif [[ ${sample} =~ "LCL7" ]]; then
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- elif [[ ${sample} =~ "LCL8" ]]; then
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
- else
- echo "only for quartet samples"
- fi
-
- cat ${sample}_SNV/summary.txt > ${project}.${sample}_SNV_precision_recall.txt
- cat ${sample}_INDEL/summary.txt > ${project}.${sample}_INDEL_precision_recall.txt
-
- >>>
-
- runtime {
- docker:docker
- cluster:cluster_config
- systemDisk:"cloud_ssd 40"
- dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
- }
-
- output {
- File rtg_vcf = "${project}.${sample}.rtg.vcf.gz"
- File rtg_vcf_index = "${project}.${sample}.rtg.vcf.gz.tbi"
- File SNV_result = "${project}.${sample}_SNV_precision_recall.txt"
- File Indel_result = "${project}.${sample}_INDEL_precision_recall.txt"
- }
- }
|