Mercurial > repos > iuc > ngsutils_bam_filter
annotate ngsutils/support/llh.py @ 2:7a68005de299 draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
author | iuc |
---|---|
date | Sun, 27 Nov 2016 15:01:21 -0500 |
parents | 4e4e4093d65d |
children |
rev | line source |
---|---|
0
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
1 ''' |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
2 Methods for calculating log-likelihoods for nucleotide frequencies |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
3 ''' |
2
7a68005de299
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents:
0
diff
changeset
|
4 import collections |
0
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
5 import math |
2
7a68005de299
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents:
0
diff
changeset
|
6 |
0
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
7 from ngsutils.support import memoize |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
8 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
9 _default_background = {'A': 0.3, 'T': 0.3, 'C': 0.2, 'G': 0.2} |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
10 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
11 NucleotideLogLikelihood = collections.namedtuple('NucleotideLogLikelihood', 'A C G T pseudo') |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
12 |
2
7a68005de299
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents:
0
diff
changeset
|
13 |
0
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
14 @memoize |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
15 def pseudo_count(N, bg): |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
16 ''' |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
17 >>> pseudo_count(100, _default_background['A']) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
18 3 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
19 >>> pseudo_count(100, _default_background['C']) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
20 2 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
21 ''' |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
22 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
23 return bg * math.sqrt(N) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
24 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
25 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
26 def calc_llh(A, C, G, T, bg=_default_background, pseudo='auto'): |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
27 if pseudo == 'auto': |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
28 N = A + C + G + T |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
29 Ap = float(A) + pseudo_count(N, bg['A']) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
30 Cp = float(C) + pseudo_count(N, bg['C']) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
31 Gp = float(G) + pseudo_count(N, bg['G']) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
32 Tp = float(T) + pseudo_count(N, bg['T']) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
33 elif pseudo: |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
34 Ap = float(A) + pseudo |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
35 Cp = float(C) + pseudo |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
36 Gp = float(G) + pseudo |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
37 Tp = float(T) + pseudo |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
38 else: |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
39 Ap = float(A) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
40 Cp = float(C) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
41 Gp = float(G) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
42 Tp = float(T) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
43 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
44 Np = Ap + Cp + Gp + Tp |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
45 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
46 freqA = Ap / Np |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
47 freqC = Cp / Np |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
48 freqG = Gp / Np |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
49 freqT = Tp / Np |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
50 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
51 return NucleotideLogLikelihood(math.log(freqA / bg['A']), math.log(freqC / bg['C']), math.log(freqG / bg['G']), math.log(freqT / bg['T']), pseudo) |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
52 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
53 |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
54 if __name__ == '__main__': |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
55 import doctest |
4e4e4093d65d
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff
changeset
|
56 doctest.testmod() |