您最多选择25个主题 主题必须以字母或数字开头,可以包含连字符 (-),并且长度不得超过35个字符

100 行
6.4KB

  1. task benchmark {
  2. File vcf
  3. File benchmarking_dir
  4. File contig
  5. File sdf
  6. String project
  7. String sample = basename(vcf,".raw.vcf.gz")
  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. gunzip ${vcf} -c > unzip.vcf
  18. cat unzip.vcf | grep '#CHROM' > name
  19. cat unzip.vcf | grep -v '#' > body
  20. cat body | grep -w '^chr1\|^chr2\|^chr3\|^chr4\|^chr5\|^chr6\|^chr7\|^chr8\|^chr9\|^chr10\|^chr11\|^chr12\|^chr13\|^chr14\|^chr15\|^chr16\|^chr17\|^chr18\|^chr19\|^chr20\|^chr21\|^chr22\|^chrX' > body.filtered
  21. if grep -q PASS "body.filtered"; then
  22. cat body.filtered | grep -v '0/0' | grep 'PASS' > body.filtered.gt
  23. cat unzip.vcf | grep '##' | grep -v 'contig' | cat - ${contig} name body.filtered.gt > filtered.vcf
  24. else
  25. cat body.filtered | grep -v '0/0' > body.filtered.gt
  26. cat unzip.vcf | grep '##' | grep -v 'contig' | cat - ${contig} name body.filtered.gt > filtered.vcf
  27. fi
  28. echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL5" > LCL5_name
  29. echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL6" > LCL6_name
  30. echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL7" > LCL7_name
  31. echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL8" > LCL8_name
  32. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip filtered.vcf -c > ${project}.${sample}.rtg.vcf.gz
  33. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.rtg.vcf.gz
  34. /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
  35. /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
  36. if [[ ${sample} =~ "LCL5" ]];then
  37. /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
  38. /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
  39. cat filtered.vcf | grep '##' > header
  40. cat filtered.vcf | grep -v '#' > body
  41. cat header LCL5_name body > LCL5.vcf
  42. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL5.vcf -c > ${project}.${sample}.reformed.vcf.gz
  43. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.reformed.vcf.gz
  44. elif [[ ${sample} =~ "LCL6" ]]; then
  45. /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
  46. /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
  47. cat filtered.vcf | grep '##' > header
  48. cat filtered.vcf | grep -v '#' > body
  49. cat header LCL6_name body > LCL6.vcf
  50. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL6.vcf -c > ${project}.${sample}.reformed.vcf.gz
  51. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.reformed.vcf.gz
  52. elif [[ ${sample} =~ "LCL7" ]]; then
  53. /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
  54. /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
  55. cat filtered.vcf | grep '##' > header
  56. cat filtered.vcf | grep -v '#' > body
  57. cat header LCL7_name body > LCL7.vcf
  58. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL7.vcf -c > ${project}.${sample}.reformed.vcf.gz
  59. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.reformed.vcf.gz
  60. elif [[ ${sample} =~ "LCL8" ]]; then
  61. /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
  62. /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
  63. cat filtered.vcf | grep '##' > header
  64. cat filtered.vcf | grep -v '#' > body
  65. cat header LCL8_name body > LCL8.vcf
  66. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL8.vcf -c > ${project}.${sample}.reformed.vcf.gz
  67. /opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.reformed.vcf.gz
  68. else
  69. echo "only for quartet samples"
  70. fi
  71. cat ${sample}_SNV/summary.txt > ${project}.${sample}_SNV_precision_recall.txt
  72. cat ${sample}_INDEL/summary.txt > ${project}.${sample}_INDEL_precision_recall.txt
  73. >>>
  74. runtime {
  75. docker:docker
  76. cluster:cluster_config
  77. systemDisk:"cloud_ssd 40"
  78. dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
  79. }
  80. output {
  81. File rtg_vcf = "${project}.${sample}.reformed.vcf.gz"
  82. File rtg_vcf_index = "${project}.${sample}.reformed.vcf.gz.tbi"
  83. File SNV_result = "${project}.${sample}_SNV_precision_recall.txt"
  84. File Indel_result = "${project}.${sample}_INDEL_precision_recall.txt"
  85. }
  86. }