annotate kmersvm/scripts/nullseq_generate.py @ 5:f99b5099ea55 draft

Uploaded
author test-svm
date Sun, 05 Aug 2012 16:50:57 -0400
parents 66088269713e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
1 import os
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
2 import os.path
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
3 import sys
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
4 import random
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
5 import optparse
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
6
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
7 from bitarray import bitarray
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
8
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
9 def read_bed_file(filename):
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
10 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
11 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
12 f = open(filename, 'r')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
13 lines = f.readlines()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
14 f.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
15
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
16 positions = []
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
17
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
18 for line in lines:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
19 if line[0] == '#':
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
20 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
21
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
22 l = line.split()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
23
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
24 positions.append((l[0], int(l[1]), int(l[2])))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
25
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
26 return positions
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
27
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
28
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
29 def bitarray_fromfile(filename):
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
30 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
31 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
32 fh = open(filename, 'rb')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
33 bits = bitarray()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
34 bits.fromfile(fh)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
35
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
36 return bits, fh
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
37
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
38
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
39 def make_profile(positions, buildname, basedir):
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
40 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
41 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
42 chrnames = sorted(set(map(lambda p: p[0], positions)))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
43
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
44 profiles = []
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
45 for chrom in chrnames:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
46 idxf_gc = os.path.join(basedir, '.'.join([buildname, chrom, 'gc', 'out']))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
47 idxf_rpt = os.path.join(basedir, '.'.join([buildname, chrom, 'rpt', 'out']))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
48
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
49 #if os.path.exists(idxf_gc) == False or os.path.exists(idxf_rpt) == False:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
50 # continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
51 bits_gc, gcf = bitarray_fromfile(idxf_gc)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
52 bits_rpt, rptf = bitarray_fromfile(idxf_rpt)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
53
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
54 for pos in positions:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
55 if pos[0] != chrom:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
56 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
57
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
58 seqid = pos[0] + ':' + str(pos[1]+1) + '-' + str(pos[2])
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
59 slen = pos[2]-pos[1]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
60 gc = bits_gc[pos[1]:pos[2]].count(True)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
61 rpt = bits_rpt[pos[1]:pos[2]].count(True)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
62
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
63 profiles.append((seqid, slen, gc, rpt))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
64
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
65 gcf.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
66 rptf.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
67
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
68 return profiles
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
69
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
70
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
71 def sample_sequences(positions, buildname, basedir, options):
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
72 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
73 """
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
74 rpt_err = options.rpt_err
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
75 gc_err = options.gc_err
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
76 max_trys = options.max_trys
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
77 norpt = options.norpt
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
78 nogc = options.nogc
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
79
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
80 chrnames = sorted(set(map(lambda p: p[0], positions)))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
81 profiles = make_profile(positions, buildname, basedir)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
82
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
83 excluded = []
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
84 if options.excludefile:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
85 excluded = read_bed_file(options.excludefile)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
86
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
87 #truncate (clear) file
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
88 f = open(options.output, 'w')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
89 f.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
90
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
91 for chrom in chrnames:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
92 if options.quiet == False:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
93 sys.stderr.write("sampling from " + chrom + "\n")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
94
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
95 idxf_na = os.path.join(basedir, '.'.join([buildname, chrom, 'na', 'out']))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
96 idxf_gc = os.path.join(basedir, '.'.join([buildname, chrom, 'gc', 'out']))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
97 idxf_rpt = os.path.join(basedir, '.'.join([buildname, chrom, 'rpt', 'out']))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
98
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
99 #this bit array is used to mark positions that are excluded from sampling
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
100 #this will be updated as we sample more sequences in order to prevent sampled sequences from overlapping
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
101 bits_na, naf = bitarray_fromfile(idxf_na)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
102 bits_gc, gcf = bitarray_fromfile(idxf_gc)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
103 bits_rpt, rptf = bitarray_fromfile(idxf_rpt)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
104
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
105 #mark excluded regions
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
106 for pos in excluded:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
107 if pos[0] != chrom:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
108 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
109 bits_na[pos[1]:pos[2]] = True
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
110
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
111 npos = 0
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
112 #mark positive regions
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
113 for pos in positions:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
114 if pos[0] != chrom:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
115 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
116 bits_na[pos[1]:pos[2]] = True
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
117 npos+=1
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
118
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
119 if options.count == 0:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
120 count = options.fold*npos
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
121 else:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
122 count = options.count
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
123
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
124 sampled_positions = []
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
125 while len(sampled_positions) < count:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
126 sampled_prof = random.choice(profiles)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
127 sampled_len = sampled_prof[1]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
128 sampled_gc = sampled_prof[2]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
129 sampled_rpt = sampled_prof[3]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
130
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
131 rpt_err_allowed = int(rpt_err*sampled_len)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
132 gc_err_allowed = int(gc_err*sampled_len)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
133 trys = 0
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
134 while trys < max_trys:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
135 trys += 1
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
136
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
137 pos = random.randint(1, bits_na.length() - sampled_len)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
138 pos_e = pos+sampled_len
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
139
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
140 if bits_na[pos:pos_e].any():
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
141 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
142
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
143 if not norpt:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
144 pos_rpt = bits_rpt[pos:pos_e].count(True)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
145 if abs(sampled_rpt - pos_rpt) > rpt_err_allowed:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
146 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
147
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
148 if not nogc:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
149 pos_gc = bits_gc[pos:pos_e].count(True)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
150 if abs(sampled_gc - pos_gc) > gc_err_allowed:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
151 continue
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
152
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
153 #accept the sampled position
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
154 #mark the sampled regions
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
155 bits_na[pos:pos_e] = True
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
156
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
157 sampled_positions.append((chrom, pos, pos_e))
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
158
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
159 #print trys, chrom, pos, pos_e, sampled_len, pos_rpt, sampled_rpt, pos_gc, sampled_gc
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
160 break
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
161 else:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
162 if options.quiet == False:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
163 sys.stderr.write(' '.join(["fail to sample from", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
164 "len=", str(sampled_len), \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
165 "rpt=", str(sampled_rpt), \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
166 "gc=", str(sampled_gc)]) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
167
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
168 naf.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
169 gcf.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
170 rptf.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
171
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
172 f = open(options.output, 'a')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
173 for spos in sorted(sampled_positions, key=lambda s: s[1]):
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
174 f.write('\t'.join([spos[0], str(spos[1]), str(spos[2])]) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
175 f.close()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
176
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
177
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
178 def main(argv=sys.argv):
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
179 usage = "usage: %prog [options] <Input Bed File> <Genome Build Name> <Base Directory>"
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
180
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
181 desc = "generate null sequences"
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
182 parser = optparse.OptionParser(usage=usage, description=desc)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
183
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
184 parser.add_option("-x", dest="fold", type="int", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
185 default = 1, help="number of sequence to sample, FOLD times of given dataset (default=1)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
186
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
187 parser.add_option("-c", dest="count", type="int", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
188 default=0, help="number of sequences to sample, override -x option (default=NA, obsolete)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
189
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
190 parser.add_option("-r", dest="rseed", type="int", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
191 default=1, help="random number seed (default=1)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
192
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
193 parser.add_option("-g", dest="gc_err", type="float", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
194 default=0.02, help="GC errors allowed (default=0.02)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
195
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
196 parser.add_option("-t", dest="rpt_err", type="float", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
197 default=0.02, help="RPT errors allowed (default=0.02)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
198
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
199 parser.add_option("-e", dest="excludefile", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
200 default="", help="filename that contains regions to be excluded (default=NA)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
201
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
202 parser.add_option("-G", dest="nogc", action="store_true", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
203 default=False, help="do not match gc-contents")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
204
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
205 parser.add_option("-R", dest="norpt", action="store_true", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
206 default=False, help="do not match repeats")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
207
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
208 parser.add_option("-m", dest="max_trys", type="int", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
209 default=10000, help="number of maximum trys to sample of one sequence (default=10000)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
210
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
211 parser.add_option("-o", dest="output", default="nullseq_output.bed", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
212 help="set the name of output file (default=nullseq_output.bed)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
213
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
214 parser.add_option("-q", dest="quiet", default=False, action="store_true", \
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
215 help="supress messages (default=false)")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
216
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
217 (options, args) = parser.parse_args()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
218
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
219 if len(args) == 0:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
220 parser.print_help()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
221 sys.exit(0)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
222
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
223 if len(args) != 3:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
224 parser.error("incorrect number of arguments")
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
225 parser.print_help()
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
226 sys.exit(0)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
227
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
228
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
229 posfile = args[0]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
230 buildname = args[1]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
231 basedir = args[2]
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
232
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
233 random.seed(options.rseed)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
234
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
235 if options.quiet == False:
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
236 sys.stderr.write('Options:\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
237 sys.stderr.write(' N fold: ' + str(options.fold) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
238 sys.stderr.write(' GC match: ' + str(not options.nogc) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
239 sys.stderr.write(' repeat match: ' + str(not options.norpt) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
240 sys.stderr.write(' GC error allowed: ' + str(options.gc_err) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
241 sys.stderr.write(' repeat error allowed: ' + str(options.rpt_err) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
242 sys.stderr.write(' random seed: ' + str(options.rseed) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
243 sys.stderr.write(' max trys: ' + str(options.max_trys) + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
244 sys.stderr.write(' excluded regions: ' + options.excludefile+ '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
245 sys.stderr.write(' output: ' + options.output + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
246 sys.stderr.write('\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
247
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
248 sys.stderr.write('Input args:\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
249 sys.stderr.write(' input bed file: ' + posfile+ '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
250 sys.stderr.write(' genome build name: ' + buildname+ '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
251 sys.stderr.write(' basedir: ' + basedir + '\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
252 sys.stderr.write('\n')
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
253
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
254 positions = read_bed_file(posfile)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
255
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
256 sample_sequences(positions, buildname, basedir, options)
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
257
66088269713e Uploaded all files tracked by git
test-svm
parents:
diff changeset
258 if __name__ == "__main__": main()