Nie możesz wybrać więcej, niż 25 tematów Tematy muszą się zaczynać od litery lub cyfry, mogą zawierać myślniki ('-') i mogą mieć do 35 znaków.

65 lines
4.2KB

  1. task benchmark {
  2. File vcf
  3. File benchmarking_dir
  4. File sdf
  5. String project
  6. String sample = basename(vcf,".raw.vcf.gz")
  7. String fasta
  8. String docker
  9. String cluster_config
  10. String disk_size
  11. command <<<
  12. set -o pipefail
  13. set -e
  14. nt=$(nproc)
  15. mkdir -p /cromwell_root/tmp
  16. cp -r ${benchmarking_dir} /cromwell_root/tmp/
  17. ls /cromwell_root/tmp/reference_datasets_v202103/ > files.txt
  18. gunzip ${vcf} -c > unzip.vcf
  19. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip unzip.vcf -c > ${project}.${sample}.rtg.vcf.gz
  20. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.rtg.vcf.gz
  21. /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
  22. /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
  23. if [[ ${sample} =~ "LCL5" ]];then
  24. /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
  25. /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
  26. elif [[ ${sample} =~ "LCL6" ]]; then
  27. /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
  28. /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
  29. elif [[ ${sample} =~ "LCL7" ]]; then
  30. /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
  31. /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
  32. elif [[ ${sample} =~ "LCL8" ]]; then
  33. /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
  34. /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
  35. else
  36. echo "only for quartet samples"
  37. fi
  38. cat ${sample}_SNV/summary.txt > ${project}.${sample}_SNV_precision_recall.txt
  39. cat ${sample}_INDEL/summary.txt > ${project}.${sample}_INDEL_precision_recall.txt
  40. >>>
  41. runtime {
  42. docker:docker
  43. cluster:cluster_config
  44. systemDisk:"cloud_ssd 40"
  45. dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
  46. }
  47. output {
  48. File rtg_vcf = "${project}.${sample}.rtg.vcf.gz"
  49. File rtg_vcf_index = "${project}.${sample}.rtg.vcf.gz.tbi"
  50. File SNV_result = "${project}.${sample}_SNV_precision_recall.txt"
  51. File Indel_result = "${project}.${sample}_INDEL_precision_recall.txt"
  52. File file_list = "files.txt"
  53. }
  54. }