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

Uploaded
author test-svm
date Sun, 05 Aug 2012 16:50:57 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
1 #!/usr/bin/env python2.7
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
2
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
3 import os
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
4 import os.path
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
5 import sys
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
6 import random
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
7 import optparse
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
8
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
9 from bitarray import bitarray
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
10
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
11 def read_bed_file(filename):
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
12 """
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
13 """
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
14 f = open(filename, 'r')
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
15 lines = f.readlines()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
16 f.close()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
17
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
18 positions = []
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
19
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
20 for line in lines:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
21 if line[0] == '#':
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
22 continue
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
23
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
24 l = line.split()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
25
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
26 positions.append((l[0], int(l[1]), int(l[2])))
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
27
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
28 return positions
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
29
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
30
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
31 def bitarray_fromfile(filename):
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
32 """
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
33 """
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
34 fh = open(filename, 'rb')
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
35 bits = bitarray()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
36 bits.fromfile(fh)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
37
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
38 return bits, fh
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
39
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
40 def get_seqid(buildname, pos):
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
41 return '_'.join( [buildname, pos[0], str(pos[1]), str(pos[2]), '+'] )
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
42
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
43 def make_profile(positions, buildname, basedir):
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
44 """
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
45 """
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
46 chrnames = sorted(set(map(lambda p: p[0], positions)))
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
47
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
48 profiles = {}
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
49 for chrom in chrnames:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
50 idxf_gc = os.path.join(basedir, '.'.join([buildname, chrom, 'gc', 'out']))
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
51 idxf_rpt = os.path.join(basedir, '.'.join([buildname, chrom, 'rpt', 'out']))
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
52
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
53 #if os.path.exists(idxf_gc) == False or os.path.exists(idxf_rpt) == False:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
54 # continue
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
55 bits_gc, gcf = bitarray_fromfile(idxf_gc)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
56 bits_rpt, rptf = bitarray_fromfile(idxf_rpt)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
57
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
58 for pos in positions:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
59 if pos[0] != chrom:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
60 continue
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
61
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
62 seqid = get_seqid(buildname, pos)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
63 slen = pos[2]-pos[1]
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
64 gc = bits_gc[pos[1]:pos[2]].count(True)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
65 rpt = bits_rpt[pos[1]:pos[2]].count(True)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
66
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
67 profiles[seqid] = (slen, gc, rpt)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
68
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
69 gcf.close()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
70 rptf.close()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
71
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
72 return profiles
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
73
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
74 def main():
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
75 parser = optparse.OptionParser()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
76 if len(sys.argv) != 5:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
77 print "Usage:", sys.argv[0], "BEDFILE BUILDNAME BASE_DIR OUT_FILE"
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
78 sys.exit()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
79 parser.add_option("-o", dest="output", default="seq_profile.txt")
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
80 (options,args) = parser.parse_args()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
81
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
82 bedfile = sys.argv[1]
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
83 buildname = sys.argv[2]
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
84 basedir = sys.argv[3]
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
85 output = options.output
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
86
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
87 positions = read_bed_file(bedfile)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
88 seqids = []
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
89 for pos in positions:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
90 seqids.append(get_seqid(buildname, pos))
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
91
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
92 profiles = make_profile(positions, buildname, basedir)
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
93
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
94 f = open(output, 'w')
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
95 f.write("\t".join(["#seq_id", "length", "GC content", "repeat fraction"]) + '\n')
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
96 for seqid in seqids:
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
97 prof = profiles[seqid]
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
98 f.write('\t'.join( map(str, [seqid, prof[0], prof[1]/float(prof[0]), prof[2]/float(prof[0])]) ) + '\n')
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
99 f.close()
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
100
f99b5099ea55 Uploaded
test-svm
parents:
diff changeset
101 if __name__ == "__main__": main()