renluyao
/
quartet-wes-germline-data-analysis-qc
Suivre 1
Ajouter aux favoris 0
Bifurcation 0
Code Tickets 0 Demandes d'ajout 0 Versions 0 Wiki Activité
Vous ne pouvez pas sélectionner plus de 25 sujets Les noms de sujets doivent commencer par une lettre ou un nombre, peuvent contenir des tirets ('-') et peuvent comporter jusqu'à 35 caractères.
6 Révisions
1 Branche
Branche: master
Branches Tags
${ item.name }
Créer la branche ${ searchTerm }
de 'master'
${ noResults }
quartet-wes-germline-data-a.../codescripts/hap_summary.py

50 lignes
3.8KB
Brut Lien permanent Annotations Historique 

  1. import pandas as pd

  2. import sys, argparse, os

  3. 

  4. parser = argparse.ArgumentParser(description="This script is to get information from hap")

  5. parser.add_argument('-hap', '--happy', type=str, help='hap.py table',  required=True)

  6. parser.add_argument('-name', '--name', type=str, help='sample name',  required=True)

  7. 

  8. args = parser.parse_args()

  9. 

  10. hap_file = args.happy

  11. name = args.name

  12. 

  13. dat = pd.read_table(hap_file)

  14. dat['QUERY.TP'] = dat['QUERY.TOTAL'].astype(int) - dat['QUERY.UNK'].astype(int) - dat['QUERY.FP'].astype(int)

  15. dat['QUERY'] = dat['QUERY.TOTAL'].astype(int) - dat['QUERY.UNK'].astype(int)

  16. indel = dat[['INDEL' in s for s in dat['Type']]]

  17. snv = dat[['SNP' in s for s in dat['Type']]]

  18. indel.reset_index(drop=True, inplace=True)

  19. snv.reset_index(drop=True, inplace=True)

  20. benchmark = pd.concat([snv, indel], axis=1)

  21. benchmark = benchmark[[ 'QUERY.TOTAL', 'QUERY','QUERY.TP','QUERY.FP','TRUTH.FN','METRIC.Precision', 'METRIC.Recall','METRIC.F1_Score']]

  22. benchmark.columns = ['SNV number','INDEL number','SNV query','INDEL query','SNV TP','INDEL TP','SNV FP','INDEL FP','SNV FN','INDEL FN','SNV precision','INDEL precision','SNV recall','INDEL recall','SNV F1','INDEL F1']

  23. benchmark = benchmark[['SNV number','INDEL number','SNV query','INDEL query','SNV TP','INDEL TP','SNV FP','INDEL FP','SNV FN','INDEL FN','SNV precision','INDEL precision','SNV recall','INDEL recall','SNV F1','INDEL F1']]

  24. benchmark['SNV precision'] = benchmark['SNV precision'].astype(float)

  25. benchmark['INDEL precision'] = benchmark['INDEL precision'].astype(float)

  26. benchmark['SNV recall'] = benchmark['SNV recall'].astype(float)

  27. benchmark['INDEL recall'] = benchmark['INDEL recall'].astype(float)

  28. benchmark['SNV F1'] = benchmark['SNV F1'].astype(float)

  29. benchmark['INDEL F1'] = benchmark['INDEL F1'].astype(float)

  30. benchmark = benchmark.round(2)

  31. 

  32. name_array = name.split("_")

  33. LCL5_1 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL5_1" + "_" + name_array[5]

  34. LCL5_2 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL5_2" + "_" + name_array[5]

  35. LCL5_3 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL5_3" + "_" + name_array[5]

  36. 

  37. LCL6_1 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL6_1" + "_" + name_array[5]

  38. LCL6_2 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL6_2" + "_" + name_array[5]

  39. LCL6_3 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL6_3" + "_" + name_array[5]

  40. 

  41. LCL7_1 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL7_1" + "_" + name_array[5]

  42. LCL7_2 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL7_2" + "_" + name_array[5]

  43. LCL7_3 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL7_3" + "_" + name_array[5]

  44. 

  45. LCL8_1 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL8_1" + "_" + name_array[5]

  46. LCL8_2 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL8_2" + "_" + name_array[5]

  47. LCL8_3 = name_array[0] + "_" + name_array[1] + "_" + name_array[2] + "_" + name_array[3] + "_"  + name_array[4] + "_" + "LCL8_3" + "_" + name_array[5]

  48. 

  49. benchmark.insert(loc=0, column='Sample', value=[LCL5_1,LCL5_2,LCL5_3,LCL6_1,LCL6_2,LCL6_3,LCL7_1,LCL7_2,LCL7_3,LCL8_1,LCL8_2,LCL8_3])

  50. benchmark.to_csv('variants.calling.qc.txt',sep="\t",index=0)