-
Notifications
You must be signed in to change notification settings - Fork 0
/
find_mature_in_precursor.py
executable file
·256 lines (203 loc) · 7.39 KB
/
find_mature_in_precursor.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
#!/usr/bin/python
'''
Created on Nov 30, 2011
@author: Chris
'''
import os
import sys
import math
import time
import copy
import random
from optparse import OptionParser
import make_svm_data as make_d
LIBSVM_PATH = '/home/schudoma/tools/libsvm-3.1/python'
if not os.path.exists(LIBSVM_PATH):
sys.stderr.write('Missing LIBSVM_PATH. Aborting.\n')
sys.exit(1)
sys.path.append(LIBSVM_PATH)
import svmutil
import svm
C_RANGE = -5, 15, 2
GAMMA_RANGE = 3, -15, -2
N_RUNS = 1000
TIMESTAMP = ''
KERNEL_TYPE = 'LINEAR'
RANDOMIZE_DATA = False
SET1 = None
SET2 = None
###
def init(argv):
global TIMESTAMP
TIMESTAMP = time.strftime('%Y-%m-%d-%H-%M-%S', time.localtime())
parser = OptionParser()
parser.add_option('-k', '--kernel', dest='kernel_type')
parser.add_option('-n', type='int', dest='n_runs')
parser.add_option('--random', dest='randomize_data', action='store_true')
parser.add_option('--s1', dest='set_1')
parser.add_option('--s2', dest='set_2')
options, args = parser.parse_args(argv)
global KERNEL_TYPE
if options.kernel_type is not None:
KERNEL_TYPE = options.kernel_type
global RANDOMIZE_DATA
if options.randomize_data is not None:
RANDOMIZE_DATA = options.randomize_data
global N_RUNS
if options.n_runs is not None:
N_RUNS = options.n_runs
global SET1
if not options.set_1 is None:
SET1 = [float(v) for v in options.set_1.split(':')]
global SET2
if not options.set_2 is None:
SET2 = [float(v) for v in options.set_2.split(':')]
return None
###
def grid_search(y, x, param, grid, cv_func, n, c_range, gamma_range):
cstart, cend, cstep = c_range
gstart, gend, gstep = gamma_range
# t0 = time.clock()
for c in xrange(cstart, cend + 1, cstep):
param.C = 2.0 ** c
# i_gamma = 0
for gamma in xrange(gstart, gend - 1, gstep):
# print 'IGAMMA', i_gamma
# i_gamma += 1
param.gamma = 2.0 ** gamma # <- Could that be a problem for linear kernel?
key = (c, gamma)
# t00 = time.clock()
grid[key] = grid.get(key, []) + cv_func(y, x, param, n=n)
# t11 = time.clock()
# print 'Search took', t11 - t00, 'seconds.'
# t1 = time.clock()
# print 'Time:', t1 - t0
return grid
###
def leave_one_out(y, x, param, n='DUMMY'):
results = []
for i, test in enumerate(zip(y, x)):
training_y = y[:i] + y[i+1:]
training_x = x[:i] + x[i+1:]
problem = svm.svm_problem(training_y, training_x)
# t0 = time.clock()
model = svmutil.svm_train(problem, param, '-q')
# t1 = time.clock()
# print 'Training took', t1 - t0, 'seconds.'
result = svmutil.svm_predict(y[i:i+1], x[i:i+1], model, '-b 1')
results.append(result + (test[0], make_d.decode(x[i], make_d.decode_dic)))
return results
###
def compute_accuracy(results):
return sum(map(float, map(lambda x: x[0]==x[1], results)))/len(results)
###
def main(argv):
global C_RANGE
global GAMMA_RANGE
global SET1
global SET2
i = 0
param_grid = {}
results = []
sum_acc = 0
init(argv[2:])
print SET1, SET2
param = svm.svm_parameter('-b 1')
if KERNEL_TYPE == 'LINEAR':
param.kernel_type = svm.LINEAR
GAMMA_RANGE = 1, 0, -2
else:
param.kernel_type = svm.RBF
cvfunc = leave_one_out
n_cv = None
use_sets = not SET1 is None and not SET2 is None
fn = argv[0]
dataset = make_d.read_data(open(fn))
data = make_d.assign_classes(dataset)
data = [(d[0], d[1][1:]) for d in data]
data = make_d.prepare_data(data)
""" Next line is just for testing. """
data = {1.0: data[1.0], 0.0: data[0.0]}
print data.keys(), [len(v) for v in data.values()]
testdata = make_d.read_data(open(argv[1]))
testset = make_d.assign_classes(testdata)
testset = [(d[0], d[1][1:]) for d in testset]
testset = make_d.prepare_data(testset)
precursor = {}
for k, v in testdata.items():
v = v[1:]
precursor[v] = precursor.get(v, []) + [int(k.split('_')[-1])]
print precursor
outfile = os.path.basename(fn)
outfile = outfile.replace('.fasta', '')
outfile = outfile.replace('.fas', '')
if use_sets:
outfile = ''.join(map(str, map(int, SET1))) + 'vs' + ''.join(map(str, map(int, SET2)))
log_name = '%s-%s-%i-%s.csv' % (TIMESTAMP,
KERNEL_TYPE,
int(RANDOMIZE_DATA),
outfile)
logfile = open(log_name, 'w')
""" Prepare test set (precursor fragments). """
testset[-1.0] = copy.deepcopy(testset[0.0])
del testset[0.0]
testset = make_d.make_set(testset, balanced_set=False, training_fraction=1.0)
""" 'Training' and 'Test' sets flipped """
test_y, test_x = testset[:2]
encoded_x = [make_d.encode(x, make_d.encode_dic) for x in test_x]
# logfile.write(',%s\n' % ','.join(map(str, map(int, test_y))))
""" Train and predict """
row = [0.0 for x in test_x]
while i < N_RUNS:
sys.stdout.write('%i ' % i)
sys.stdout.flush()
set1 = dict([item for item in data.items()
if item[0] == 1.0])
set2 = dict([item for item in data.items()
if item[0] == 0.0])
set1 = make_d.make_set(set1, training_fraction=1.0)
set2 = make_d.make_set(set2, training_fraction=1.0)
new_sets = {1.0: set1[1], -1.0: set2[1]}
sets = make_d.make_set(new_sets, training_fraction=1.0)
train_y, train_x, dummy_y, dummy_x = sets
print [len(x) for x in sets]
train_x = [make_d.encode(x, make_d.encode_dic) for x in train_x]
t0 = time.clock()
param_grid = {}
param_grid = grid_search(train_y, train_x, param, param_grid,
cvfunc, n_cv, C_RANGE, GAMMA_RANGE)
t1 = time.clock()
print 'Time:', t1 - t0, 'Remaining:', (N_RUNS-(i+1)) * (t1 - t0)
ranking = []
for k, v in param_grid.items():
recognized = [v_i[0][0] == v_i[3] for v_i in v]
recog_rate = sum(map(int, recognized))/float(len(recognized))
# print k, recog_rate, len(recognized)
ranking.append((recog_rate, k))
ranking.sort()
param.C, param.gamma = map(lambda x: 2**x, ranking[-1][1])
problem = svm.svm_problem(train_y, train_x)
model = svmutil.svm_train(problem, param, '-q')
result = svmutil.svm_predict(test_y, encoded_x, model, '-b 1')
## print result
## print zip(test_y, test_x)
cur_result = zip(result[0], test_y)
for row_i, res in enumerate(result[0]):
if res == -1.0:
res = 0.0
row[row_i] += res
# logfile.write('%i,%s\n' % (i, ','.join(map(str, map(int, result[0])))))
i += 1
pass
row = map(lambda x: x/N_RUNS, row)
pos_on_precursor = []
for dat in zip(row, test_x, test_y):
#print dat
pos_on_precursor.append((precursor[dat[1]][0],) + dat)
del precursor[dat[1]][0]
for dat in sorted(pos_on_precursor):
print dat
logfile.write('%05i,%f,%s,%f\n' % dat)
logfile.close()
return None
if __name__ == '__main__': main(sys.argv[1:])