-
Notifications
You must be signed in to change notification settings - Fork 0
/
SNPfilter.py
executable file
·273 lines (250 loc) · 9.27 KB
/
SNPfilter.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
import sys
import pysam
import bamio
import numpy as np
import time
import os
from scipy.stats import fisher_exact
from sklearn import svm
from sklearn import preprocessing
from sklearn.grid_search import GridSearchCV
vcf='/biocluster/data/biobk/user_test/zhusihui/mydata/dbsnp_138.b37.vcf'
vcfidx=bamio.Tabix(vcf,chrpos=0,start=1,end=1)
def penaltyScore(altbase,queryStr):
gapflag = 0
polymerScore = 0
for i in range(len(queryStr)):
if queryStr[i] == altbase:
print i
if gapflag == 1:
polymerScore -= 2
elif gapflag == 1 and i == 0:
polymerScore -= 6
else:
polymerScore -= 1
gapflag = 1
else:
polymerScore += 2
return polymerScore
class SNPfactory(object):
def __init__(self,sample,chr,pos,refbase,altbase,totaldepth,basefreqDict,tailDistance,upbases,downbases,basequality,basestr,mapQ,):
self.sample = sample
self.chr = chr
self.pos = int(pos)
self.ref = refbase
self.alt = altbase
self.totaldepth = int(totaldepth)
self.basefreqDict = basefreqDict
self.tailDistance = float(tailDistance)
self.upbases = upbases
self.downbases = downbases
self.basequality = basequality
self.basestr = basestr
self.mapQ = mapQ
self.SBscore = 0
self.transition = 1
self.polymerScore = 0
def TiTv(self,):
# to determine a mutation is a transversion or transition; transition mutations are generated at higher frequency than transversions
refbase = self.ref
altbase = self.alt
tmpstr = refbase+altbase
transitionFlag = 1
if tmpstr in ['AG','GA','CT','TC']:
transitionFlag = 1
else: #['AT','TA','AC','CA','CG','GC','GT','TG']
transitionFlag = 0
self.transition = transitionFlag
return 0
def calSBscore(self):
majorFreq = self.basefreqDict[self.ref]
minorFreq = self.basefreqDict[self.alt]
self.majorFreq = majorFreq
self.minorFreq = minorFreq
# caculate strand bias score based GATK-SB score ,referred in Yan Guo,BMC Genomics 2012
majordepth,majorfrac,other=majorFreq.split(";")
majorFor,majorRev = other.split(",")
self.majordepth = int(majordepth)
self.majorfrac = float(majorfrac.strip("%"))/100.0
minordepth,minorfrac,other=minorFreq.split(";")
minorFor,minorRev = other.split(",")
self.minordepth = int(minordepth)
self.minorfrac = float(minorfrac.strip("%"))/100.0
majorFor=majorFor.strip("+")
majorRev=majorRev.strip("-")
minorFor=minorFor.strip("+")
minorRev=minorRev.strip("-")
majorFor,majorRev,minorFor,minorRev = map(int,[majorFor,majorRev,minorFor,minorRev])
commonDom=float(minorFor+minorRev)/(majorFor+majorRev+minorFor+minorRev)
'''
S1=minorFor*majorRev/(float(majorFor+minorFor)*(majorRev+minorRev))
S2=majorFor*minorRev/(float(majorFor+minorFor)*(majorRev+minorRev))
score=np.max([S1/commonDom,S2/commonDom])
'''
S1 = np.abs(minorFor/float(majorFor+minorFor) - minorRev/float(majorRev+minorRev))
score = S1/commonDom
score = 1 - fisher_exact(np.asarray([[majorFor,minorFor],[majorRev,minorRev]]))[1]
self.SBscore = score
return 0
def homopolymerPenalty(self):
queryStr = ''.join(reversed(list(self.upbases)))
uppolymerScore = 0
downpolymerScore = 0
altbase = self.alt
uppolymerScore = penaltyScore(altbase,queryStr)
queryStr = self.downbases
downpolymerScore = penaltyScore(altbase,queryStr)
self.polymerScore = min(uppolymerScore,downpolymerScore)
return 0
def known_novel_anno(self,vcfidx):
chr = self.chr
pos = self.pos
items = vcfidx.fetch(chr,pos,pos+1)
knownflag = 0
#print str(pos) + "\t",
try:
items.next()
knownflag = 1
except StopIteration,e:
pass
'''
items = vcfidx.fetch(chr,pos,pos+1)
for item in items:
print item
'''
self.knownflag = knownflag
return 0
def sequenceErr(self):
altbasescore = 0
altreadscore = 0
altbaseNum = 0.0
self.basequality = map(int,self.basequality.split(";"))
self.mapQ = map(int,self.mapQ.split(";"))
for i,basestr in enumerate(self.basestr):
if basestr == self.alt:
altbasescore += self.basequality[i]
altreadscore += self.mapQ[i]
altbaseNum += 1
self.altbaseAveScore = altbasescore/altbaseNum
self.altreadAveScore = altreadscore/altbaseNum
return 0
def generateData(fh,traindata=0):
backgroudDict = {}
f=file("/biocluster/data/biobk/user_test/zhusihui/SNP_classification/database_file",'r')
for line in f:
if line.startswith("#"):
continue
arr=line.rstrip("\n").split("\t")
chr,start,end,sample,ref,alt,owner,gene,panel,flagval=arr
backgroudDict["%s-%s-%s-%s-%s"%(arr[0],arr[1],arr[3],arr[4],arr[5])] = arr[-1]
f.close()
trainDatasetsX = []
trainDatasetsY = []
keystrs = []
for line in fh:
if line.startswith("#"):
continue
sample,chr,pos,refbase,altbase,totaldepth,Afreq,Cfreq,Gfreq,Tfreq,tailDistance,upbases,downbases,basequality,basestr,mapQ,owner = line.rstrip("\n").split("\t")
basefreqDict = {'A':Afreq,'C':Cfreq,'G':Gfreq,'T':Tfreq}
SNPobj = SNPfactory(sample,chr,pos,refbase,altbase,totaldepth,basefreqDict,tailDistance,upbases,downbases,basequality,basestr,mapQ)
SNPobj.TiTv()
SNPobj.calSBscore()
SNPobj.homopolymerPenalty()
SNPobj.known_novel_anno(vcfidx)
SNPobj.sequenceErr()
keystr = "\t".join([SNPobj.sample,SNPobj.chr,str(SNPobj.pos+1),SNPobj.ref,SNPobj.alt])
outstr = [SNPobj.sample,SNPobj.chr,SNPobj.pos,SNPobj.ref,SNPobj.alt,SNPobj.transition ,SNPobj.majorfrac,SNPobj.minorfrac,SNPobj.polymerScore ,SNPobj.knownflag,SNPobj.altbaseAveScore,SNPobj.altreadAveScore]
#if "%s-%s-%s-%s-%s"%(SNPobj.chr,str(SNPobj.pos+1),SNPobj.sample,SNPobj.ref,SNPobj.alt) not in backgroudDict:
# continue
if not traindata:
trainDatasetsX.append(map(float,[SNPobj.transition,SNPobj.majorfrac,SNPobj.minorfrac,SNPobj.polymerScore,SNPobj.knownflag,SNPobj.altbaseAveScore,SNPobj.altreadAveScore,SNPobj.SBscore]))
trainDatasetsY.append(0)
keystrs.append(keystr)
else:
try :
if backgroudDict["%s-%s-%s-%s-%s"%(SNPobj.chr,str(SNPobj.pos+1),SNPobj.sample,SNPobj.ref,SNPobj.alt)]=="True":
trainDatasetsY.append(1)
else:
trainDatasetsY.append(0)
trainDatasetsX.append(map(float,[SNPobj.transition,SNPobj.majorfrac,SNPobj.minorfrac,SNPobj.polymerScore,SNPobj.knownflag,SNPobj.altbaseAveScore,SNPobj.altreadAveScore,SNPobj.SBscore]))
except KeyError,e:
pass
return trainDatasetsX,trainDatasetsY,keystrs
def fmtresult(rawTabfile,keystrs,predictArr):
fh = file(rawTabfile,'r')
fout = file(rawTabfile.rsplit(".",1)[0]+"_classed_pos.xls",'w')
fouttmp = file(rawTabfile.rsplit(".",1)[0]+"_classed_neg.xls",'w')
for line in fh:
if line.startswith("#"):
fout.write(line)
continue
arr=line.rstrip("\n").split("\t")
if arr[8] != "SNP":
fout.write(line)
continue
#chr,start,samplename,ref,alt = arr[4],arr[5],arr[0],arr[9],arr[10]
subarr = [arr[0],arr[4],arr[5],arr[9],arr[10]]
if predictArr[keystrs.index("\t".join(subarr))]:
fout.write(line)
else:
fouttmp.write(line)
fout.close()
fouttmp.close()
return 0
def fmtstr(x):
return "%.3f"%x
def scaleFeature(mat,colidx,vmax,vmin,traindata=0):
# scale to [-1,1]
if traindata:
vmin = np.min(mat[:,colidx])
vmax = np.max(mat[:,colidx])
mat[:,colidx] = -1.0 + (1.0 - -1.0) * (mat[:,colidx] - vmin)/(vmax-vmin)
return mat,vmax,vmin
def __main():
start_time = time.time()
ret = 0
if len(sys.argv) != 4:
sys.stderr.write("usage:\npython SNPfilter.py sampleCollection.xls test_predict_SNP_data test.annotate.meaningful.xls\n")
sys.exit(0)
fh=file(sys.argv[1],'r')
trainDatasetsX,trainDatasetsY,keystrs = generateData(fh,traindata=1)
fh.close()
fh=file(sys.argv[2],'r')
predictDatasetsX,predictDatasetsY,keystrs = generateData(fh)
fh.close()
### classification
trainDatasetsX = np.asarray(trainDatasetsX)
predictDatasetsX = np.asarray(predictDatasetsX)
clf = svm.SVC(kernel='linear',C=1000.,)#class_weight={0: 10, 1: 10, 2: 10, 3: 30, 4: 10, 5: 30, 6: 30, 7: 30})
if len(predictDatasetsX) == 0:
sys.stderr.write("empty array found here,please check your input data first!!!\nThis program can only be applied for SNP F/T determenation for now, so it'll ignore InDel record.\n ")
ret = 1
else:
### scale data,only for basequal and mappingQual
# ref http://suanfazu.com/t/is-scaling-of-feature-values-in-libsvm-necessary/2030/2
'''
trainDatasetsX,vmax,vmin = scaleFeature(trainDatasetsX,-3,0,0,traindata=1)
predictDatasetsX,vmax,vmin = scaleFeature(predictDatasetsX,-3,vmax,vmin,traindata=0)
trainDatasetsX,vmax,vmin = scaleFeature(trainDatasetsX,-2,0,0,traindata=1)
predictDatasetsX,vmax,vmin = scaleFeature(predictDatasetsX,-2,vmax,vmin,traindata=0)
'''
min_max_scaler = preprocessing.MinMaxScaler()
trainDatasetsX = min_max_scaler.fit_transform(trainDatasetsX)
predictDatasetsX = min_max_scaler.fit_transform(predictDatasetsX)
clf.fit(trainDatasetsX, trainDatasetsY)
predictArr = clf.predict(predictDatasetsX)
f=file('%s_log.xls'%(os.path.basename(sys.argv[3]).split(".",1)[0]),'w')
f.write("#TiTv\tmajorF\tminorF\tpolymer\tknownFlag\tbaseQ\tmappingQ\tfishScore\n")
for i in range(len(predictArr)):
f.write("\t".join(map(fmtstr,predictDatasetsX[i,:].tolist()))+'\t'+str(predictArr[i])+"\n")
f.close()
fmtresult(sys.argv[3],keystrs,predictArr)
end_time = time.time()
sys.stderr.write("Task cost %ds\n"%(end_time-start_time))
return ret
if __name__ == "__main__":
ret= __main()
if ret != 0:
sys.stderr.write("Task Failed !!!\n")
else:
sys.stderr.write("Task Done ! \n ")