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
|
#!/usr/bin/python
import sys
import math
import numpy
def nCk(n,k):
return math.factorial(n) / math.factorial(k) / math.factorial(n - k)
usage = "usage: doMS.py [ --priorfile filename | --flatmodelprior | --flatphenoprior ] [ --mode 1|2|3|4 ] input.select.multinom \n"
usageArgs = usage
usageArgs += "\n"
usageArgs += "Arguments to set the analysis mode: \n"
usageArgs += " --mode 1 - Bayes factor normalization (outputs a logBF for each model relative to the null model)\n"
usageArgs += " --mode 2 - Posterior normalization (ouputs a posterior probability for each model) [requires a prior]\n"
usageArgs += " --mode 3 - Phenotype specificity summary (outputs the best phenotype-specific and shared models, with a Bayes factor in favour of the former)\n"
usageArgs += " --mode 4 - Credible set summary (outputs the best model, the posterior for this model, and a 95% credible set of models)\n"
usageArgs += "\n"
usageArgs += "Arguments to set the prior: \n"
usageArgs += " --priorfile filename - specifies a model prior file (two columns, first column is the model name and second column is the prior probability for that model) \n"
usageArgs += " --flatmodelprior - use a default uniform prior on models \n"
usageArgs += " --flatphenoprior - use a default uniform prior on phenotype counts \n"
usageArgs += "\n"
### process the arguments
c = 1
priorfile = ""
priormode = 0
mode = 0
if len(sys.argv) == 1:
sys.stderr.write(usageArgs)
sys.exit(1)
while (c < len(sys.argv) - 1):
if sys.argv[c] == "--h" or sys.argv[c] == "--help" or sys.argv[c] == "-h":
sys.stderr.write(usageArgs)
sys.exit(1)
if sys.argv[c] == "--priorfile":
c += 1
priorfile = sys.argv[c]
c += 1
continue
if sys.argv[c] == "--flatmodelprior":
c += 1
priormode = 1
continue
if sys.argv[c] == "--flatphenoprior":
c += 1
priormode = 2
continue
if sys.argv[c] == "--mode":
c+= 1
mode = int(sys.argv[c])
c += 1
continue
sys.stderr.write("Error: argument " + sys.argv[c] + " not recognised\n")
sys.stderr.write(usage)
sys.exit(1)
if c >= len(sys.argv):
sys.stderr.write("Error: too few arguments\n")
sys.stderr.write(usage)
sys.exit(1)
if mode == 1:
sys.stderr.write("Analysis mode 1 (Bayes factor normalization)\n")
elif mode == 2:
sys.stderr.write("Analysis mode 2 (posterior normalization)\n")
elif mode == 3:
sys.stderr.write("Analysis mode 3 (Phenotype specificity report)\n")
elif mode == 4:
sys.stderr.write("Analysis mode 4 (Credible set report)\n")
elif mode == 0:
sys.stderr.write("No analysis mode specified, defaulting to mode 1 (Bayes factor normalization)\n")
mode = 1
else:
sys.stderr.write("Error: analysis mode " + mode + " not recognised\n")
sys.exit(1)
if (mode == 1 or mode == 3) and (priorfile != "" or priormode != 0):
sys.stderr.write("Warning: You have specified a prior on models, but this analysis mode does not use it.\n")
if (mode == 2 or mode == 4) and (priorfile == "" and priormode == 0):
sys.stderr.write("Warning: You have not specified a prior on models, but this analysis mode requires one. Defaulting to a flat prior on models\n")
priormode = 1
if (priorfile != ""):
if priormode != 0:
sys.stderr.write("Error: You have specified a prior mode and a prior file. Please select one or the other.\n")
sys.exit(1)
sys.stderr.write("Reading prespecified prior from file " + priorfile + ".\n")
PRIORFILE = open(priorfile)
priordict = {}
for line_raw in PRIORFILE:
line = line_raw.split()
priordict[line[0]] = float(line[1])
if "NULL" not in priordict:
sys.stderr.write("Warning: No prior specified for the NULL model. Setting it to zero.\n")
priordict["NULL"] = 0.0
if priormode == 1:
sys.stderr.write("Using default prior (uniform across models)\n")
if priormode == 2:
sys.stderr.write("Using default prior (uniform across phenotype counts)\n")
FILE = open(sys.argv[c])
phenos = set([])
header = FILE.readline().split()
if priorfile != "":
priors = [priordict["NULL"]]
ModelSize = [0]
for h in header[6:]:
if priorfile != "":
if h not in priordict:
sys.stderr.write("Error: No prior specified for model " + h + "\n")
sys.exit(1)
priors.append(priordict[h])
temp = set(h.split("_")[1:])
phenos |= temp
ModelSize.append(len(temp))
Npheno = len(phenos)
Nmodel = len(header) - 5
#print Npheno, Nmodel
if Nmodel != 2**Npheno:
sys.stderr.write("Error: Wrong number of unique phenotypes in header relative to columns. Input data misformatted." + h + "\n")
sys.exit(1)
if priormode != 0:
if (priormode == 1):
priors = [1.0/Nmodel]
else:
priors = [1.0/(Npheno + 1)]
for i in range(len(header[6:])):
if (priormode == 1):
priors.append(1.0/Nmodel)
else:
k = ModelSize[i + 1]
#print k
priors.append(1.0/((Npheno + 1)*nCk(Npheno,k)))
#print priors
if mode == 1 or mode == 2:
print "\t".join(header)
if mode == 3:
print "\t".join(header[0:5]),
print "BestSpecModel\tBestSharedModel\tLogSpecificityBF"
if mode == 4:
print "\t".join(header[0:5]),
print "BestModel\tBestModelProb\tCredibleSet"
for line_raw in FILE:
line = line_raw.split()
print "\t".join(line[0:5]),
BFs = [0.0]
for i in range(6,len(line)):
BFs.append(float(line[i]) - float(line[5]))
if mode == 2 or mode == 4:
posteriors = numpy.exp(numpy.array(BFs)*numpy.array(priors))/sum(numpy.exp(numpy.array(BFs)*numpy.array(priors)))
if (mode == 1):
print "\t".join([str(x) for x in BFs])
elif (mode == 2):
print "\t".join([str(x) for x in posteriors])
elif (mode == 3):
bestShare = -1
bestShareBF = -1e100
bestUnique = -1
bestUniqueBF = -1e100
for i in range(Nmodel):
if ModelSize[i] == 1 and BFs[i] > bestUniqueBF:
bestUniqueBF = BFs[i]
bestUnique = i
if ModelSize[i] > 1 and BFs[i] > bestShareBF:
bestShareBF = BFs[i]
bestShare = i
#print bestShare, bestShareBF, bestUnique, bestUniqueBF
print header[5 + bestUnique],header[5 + bestShare], bestUniqueBF - bestShareBF
elif (mode == 4):
order = numpy.argsort(-posteriors)
CredSet = header[5 + order[0]]
CredValue = posteriors[order[0]]
i = 1
while (CredValue < 0.95 and i < Nmodel):
CredSet += "&" + header[5 + order[i]]
CredValue += posteriors[order[i]]
i += 1
print header[5 + order[0]],posteriors[order[0]],CredSet
#print phenos
#print priors
|
