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

52 行
2.2KB

  1. task quartet_mendelian {
  2. Array[File] summary
  3. Array[File] LCL5_hap
  4. Array[File] LCL6_hap
  5. Array[File] LCL7_hap
  6. Array[File] LCL8_hap
  7. String docker
  8. String project
  9. String cluster_config
  10. String disk_size
  11. command <<<
  12. cat ${sep=" " summary} | grep -v 'Family'> mendelian.summary
  13. sed '1iFamily\tTotal_Variants\tMendelian_Concordant_Variants\tMendelian_Concordance_Rate' mendelian.summary > mendelian.txt
  14. cat mendelian.txt | grep 'INDEL' | cut -f4 | grep -v 'Mendelian_Concordance_Rate' | awk '{for(i=1;i<=NF;i++) {sum[i] += $i; sumsq[i] += ($i)^2}}
  15. END {for (i=1;i<=NF;i++) {
  16. printf "%f %f \n", sum[i]/NR, sqrt((sumsq[i]-sum[i]^2/NR)/NR)}
  17. }' >> quartet_indel_aver-std.txt
  18. cat mendelian.txt | grep 'SNV' | cut -f4 | grep -v 'Mendelian_Concordance_Rate' | awk '{for(i=1;i<=NF;i++) {sum[i] += $i; sumsq[i] += ($i)^2}}
  19. END {for (i=1;i<=NF;i++) {
  20. printf "%f %f \n", sum[i]/NR, sqrt((sumsq[i]-sum[i]^2/NR)/NR)}
  21. }' >> quartet_snv_aver-std.txt
  22. cat ${sep=" " LCL5_hap} ${sep=" " LCL6_hap} ${sep=" " LCL7_hap} ${sep=" " LCL8_hap} | grep ALL | sed s'/,/\t/g' > hap.summary
  23. sed '1i\Type\tFilter\tTRUTH.TOTAL\tTRUTH.TP\tTRUTH.FN\tQUERY.TOTAL\tQUERY.FP\tQUERY.UNK\tFP.gt\tMETRIC.Recall\tMETRIC.Precision\tMETRIC.Frac_NA\tMETRIC.F1_Score\tTRUTH.TOTAL.TiTv_ratio\tQUERY.TOTAL.TiTv_ratio\tTRUTH.TOTAL.het_hom_ratio\tQUERY.TOTAL.het_hom_ratio' hap.summary > precision_recall
  24. python /opt/hap_summary.py -hap precision_recall -name ${project}
  25. cat variants.calling.qc.txt | cut -f12- | grep -v 'SNV' | awk '{for(i=1;i<=NF;i++) {sum[i] += $i; sumsq[i] += ($i)^2}}
  26. END {for (i=1;i<=NF;i++) {
  27. printf "%f %f \n", sum[i]/NR, sqrt((sumsq[i]-sum[i]^2/NR)/NR)}
  28. }' >> reference_datasets_aver-std.txt
  29. >>>
  30. runtime {
  31. docker:docker
  32. cluster:cluster_config
  33. systemDisk:"cloud_ssd 40"
  34. dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
  35. }
  36. output {
  37. File mendelian_summary = "mendelian.txt"
  38. File snv_aver_std = "quartet_snv_aver-std.txt"
  39. File indel_aver_std = "quartet_indel_aver-std.txt"
  40. File pr = "precision_recall"
  41. File hap_summary = "variants.calling.qc.txt"
  42. File precision_recall_aver_std = "reference_datasets_aver-std.txt"
  43. }
  44. }