Mercurial > repos > yutaka-saito > bsfcall
comparison bin/last-map-probs @ 0:06f8460885ff
migrate from GitHub
author | yutaka-saito |
---|---|
date | Sun, 19 Apr 2015 20:51:13 +0900 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:06f8460885ff |
---|---|
1 #! /usr/bin/env python | |
2 | |
3 # Copyright 2010, 2011, 2012, 2014 Martin C. Frith | |
4 | |
5 # Read query-genome alignments: write them along with the probability | |
6 # that each alignment is not the true mapping of its query. These | |
7 # probabilities make the risky assumption that one of the alignments | |
8 # reported for each query is correct. | |
9 | |
10 import sys, os, fileinput, math, optparse, signal | |
11 | |
12 def logsum(numbers): | |
13 """Adds numbers, in log space, to avoid overflow.""" | |
14 m = max(numbers) | |
15 s = sum(math.exp(i - m) for i in numbers) | |
16 return m + math.log(s) | |
17 | |
18 def mafScore(aLine): | |
19 for word in aLine.split(): | |
20 if word.startswith("score="): | |
21 return float(word[6:]) | |
22 raise Exception("found an alignment without a score") | |
23 | |
24 def namesAndScores(lines): | |
25 queryNames = [] | |
26 scores = [] | |
27 for line in lines: | |
28 if line[0] == "a": # faster than line.startswith("a") | |
29 s = mafScore(line) | |
30 scores.append(s) | |
31 sLineCount = 0 | |
32 elif line[0] == "s": | |
33 sLineCount += 1 | |
34 if sLineCount == 2: queryNames.append(line.split(None, 2)[1]) | |
35 elif line[0].isdigit(): # we have an alignment in tabular format | |
36 w = line.split(None, 7) | |
37 scores.append(float(w[0])) | |
38 queryNames.append(w[6]) | |
39 return queryNames, scores | |
40 | |
41 def scoreTotals(queryNames, scores, temperature): | |
42 queryLists = {} | |
43 for n, s in zip(queryNames, scores): | |
44 queryLists.setdefault(n, []).append(s / temperature) | |
45 return dict((k, logsum(v)) for k, v in queryLists.iteritems()) | |
46 | |
47 def writeOneBatch(lines, queryNames, scores, denominators, opts, temperature): | |
48 isWanted = True | |
49 i = 0 | |
50 for line in lines: | |
51 if line[0] == "a": | |
52 s = scores[i] | |
53 p = 1.0 - math.exp(s / temperature - denominators[queryNames[i]]) | |
54 i += 1 | |
55 if s < opts.score or p > opts.mismap: | |
56 isWanted = False | |
57 else: | |
58 newLineEnd = " mismap=%.3g\n" % p | |
59 line = line.rstrip() + newLineEnd | |
60 elif line[0].isdigit(): # we have an alignment in tabular format | |
61 s = scores[i] | |
62 p = 1.0 - math.exp(s / temperature - denominators[queryNames[i]]) | |
63 i += 1 | |
64 if s < opts.score or p > opts.mismap: continue | |
65 newLineEnd = "\t%.3g\n" % p | |
66 line = line.rstrip() + newLineEnd | |
67 if isWanted: print line, | |
68 if line.isspace(): isWanted = True # reset at end of maf paragraph | |
69 | |
70 def doOneBatch(lines, opts, temperature): | |
71 queryNames, scores = namesAndScores(lines) | |
72 denominators = scoreTotals(queryNames, scores, temperature) | |
73 writeOneBatch(lines, queryNames, scores, denominators, opts, temperature) | |
74 | |
75 def readHeaderOrDie(lines): | |
76 t = 0.0 | |
77 e = -1 | |
78 for line in lines: | |
79 if line.startswith("#") or line.isspace(): | |
80 print line, | |
81 for i in line.split(): | |
82 if i.startswith("t="): t = float(i[2:]) | |
83 elif i.startswith("e="): e = int(i[2:]) | |
84 if t > 0 and e >= 0: break | |
85 else: | |
86 raise Exception("I need a header with t= and e=") | |
87 return t, e | |
88 | |
89 def lastMapProbs(opts, args): | |
90 f = fileinput.input(args) | |
91 temperature, e = readHeaderOrDie(f) | |
92 if opts.score < 0: opts.score = e + round(temperature * math.log(1000)) | |
93 lines = [] | |
94 | |
95 for line in f: | |
96 if line.startswith("# batch"): | |
97 doOneBatch(lines, opts, temperature) | |
98 lines = [] | |
99 lines.append(line) | |
100 doOneBatch(lines, opts, temperature) | |
101 | |
102 if __name__ == "__main__": | |
103 signal.signal(signal.SIGPIPE, signal.SIG_DFL) # avoid silly error message | |
104 | |
105 usage = """ | |
106 %prog --help | |
107 %prog [options] lastal-alignments""" | |
108 | |
109 description = "Calculate a mismap probability for each alignment. This is the probability that the alignment does not reflect the origin of the query sequence, assuming that one reported alignment does reflect the origin of each query." | |
110 | |
111 op = optparse.OptionParser(usage=usage, description=description) | |
112 op.add_option("-m", "--mismap", type="float", default=0.01, metavar="M", | |
113 help="don't write alignments with mismap probability > M (default: %default)") | |
114 op.add_option("-s", "--score", type="float", default=-1, metavar="S", | |
115 help="don't write alignments with score < S (default: e+t*ln[1000])") | |
116 (opts, args) = op.parse_args() | |
117 if not args and sys.stdin.isatty(): | |
118 op.print_help() | |
119 op.exit() | |
120 | |
121 try: lastMapProbs(opts, args) | |
122 except KeyboardInterrupt: pass # avoid silly error message | |
123 except Exception, e: | |
124 prog = os.path.basename(sys.argv[0]) | |
125 sys.exit(prog + ": error: " + str(e)) |