Mercurial > repos > melissacline > ucsc_cancer_utilities

comparison mergeXenaMutation.py @ 55:1093078e7976

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
merge mutation data conform to new mutationVector data standard
author jingchunzhu
date Fri, 18 Sep 2015 10:24:39 -0700
parents 9806198df91f
children
comparison
equal deleted inserted replaced
54:59dbe857f5d4 55:1093078e7976
1 #!/usr/bin/env python 1 #!/usr/bin/env python
2 2
3 import argparse 3 import argparse
4 import string, os, sys 4 import string, os, sys
5 5
6 requiredCOLs = ["chr", "start","end","reference","alt","gene","effect"] 6 requiredCOLs = ["chr", "start","end","reference","alt"]
7 7
8 def headerError(filename, column, ferror): 8 def headerError(filename, column, ferror):
9 ferror.write(filename +" does not have column " + column+"\n") 9 ferror.write(filename +" does not have column " + column+"\n")
10 ferror.close() 10 ferror.close()
11 sys.exit(1) 11 sys.exit(1)
24 #header 24 #header
25 line = fin.readline() 25 line = fin.readline()
26 fin.close() 26 fin.close()
27 if line [0]=="#": 27 if line [0]=="#":
28 line = line[1:-1] 28 line = line[1:-1]
29 data = string.split(line,"\t") 29 data = string.split(string.strip(line),"\t")
30 30
31 columnDic["chr"]= findAnyValueInList (["chr","chrom"], data) 31 columnDic["chr"]= findAnyValueInList (["chr","chrom", "Chr"], data)
32 if columnDic["chr"] ==-1: 32 if columnDic["chr"] ==-1:
33 headerError(infile, "chr", ferror) 33 headerError(infile, "chr", ferror)
34 34
35 columnDic["start"]= findAnyValueInList (["start","chrStart"], data) 35 columnDic["start"]= findAnyValueInList (["start","chrStart","Start"], data)
36 if columnDic["start"] == -1: 36 if columnDic["start"] == -1:
37 headerError(infile, "start", ferror) 37 headerError(infile, "start", ferror)
38 38
39 columnDic["end"]= findAnyValueInList (["end","chrEnd"], data) 39 columnDic["end"]= findAnyValueInList (["end","chrEnd", "End"], data)
40 if columnDic["end"] == -1: 40 if columnDic["end"] == -1:
41 headerError(infile, "end", ferror) 41 headerError(infile, "end", ferror)
42 42
43 columnDic["alt"]= findAnyValueInList (["alt"], data) 43 columnDic["alt"]= findAnyValueInList (["alt","Alt"], data)
44 if columnDic["alt"] == -1: 44 if columnDic["alt"] == -1:
45 headerError(infile, "alt", ferror) 45 headerError(infile, "alt", ferror)
46 46
47 columnDic["reference"]= findAnyValueInList (["reference","ref"], data) 47 columnDic["reference"]= findAnyValueInList (["reference","ref","Reference","Ref"], data)
48 if columnDic["reference"] == -1: 48 if columnDic["reference"] == -1:
49 headerError(infile, "reference", ferror) 49 headerError(infile, "reference", ferror)
50 50
51 columnDic["gene"]= findAnyValueInList (["gene"], data) 51 #columnDic["gene"]= findAnyValueInList (["gene","Gene"], data)
52 if columnDic["gene"] == -1: 52 #if columnDic["gene"] == -1:
53 headerError(infile, "gene", ferror) 53 # headerError(infile, "gene", ferror)
54 54
55 columnDic["effect"]= findAnyValueInList (["effect"], data) 55 #columnDic["effect"]= findAnyValueInList (["effect"], data)
56 if columnDic["effect"] == -1: 56 #if columnDic["effect"] == -1:
57 headerError(infile, "effect", ferror) 57 # headerError(infile, "effect", ferror)
58 58
59 requiredCols = columnDic.keys() 59 requiredCols = columnDic.keys()
60 requiredColsPos = columnDic.values() 60 requiredColsPos = columnDic.values()
61 for i in range(1,len(data)): 61 for i in range(1,len(data)):
62 if i not in requiredColsPos: 62 if i not in requiredColsPos:
72 if col not in allCols: 72 if col not in allCols:
73 allCols.append(col) 73 allCols.append(col)
74 return 74 return
75 75
76 def outputHeader (requiredCOLs,allCols,fout): 76 def outputHeader (requiredCOLs,allCols,fout):
77 fout.write("#sample") 77 fout.write("sample")
78 for col in requiredCOLs: 78 for col in requiredCOLs:
79 fout.write("\t"+col) 79 fout.write("\t"+col)
80 for col in allCols: 80 for col in allCols:
81 if col not in requiredCOLs: 81 if col not in requiredCOLs:
82 fout.write("\t"+col) 82 fout.write("\t"+col)