annotate annotateVCF.py @ 6:22f6a0e7424f draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
author iuc
date Fri, 16 Oct 2020 17:04:35 +0000
parents 84dcccebad3d
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
1 #!/usr/bin/env python3
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
2
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
3 # Takes in VCF file and a samtools mpileup output file
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
4 # Fills in annotation for the VCF file including AF, DP
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
5 # SB, and DP4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
6 #
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
7 # Usage statement:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
8 # python annotateVCF.py in_vcf.vcf in_mpileup.txt out_vcf.vcf
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
9 #
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
10 # Can generate in_mileup.txt with samtools mpileup (and can restrict which sites to generate pileups for with in_vcf.vcf)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
11
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
12 # 08/24/2020 - Nathan P. Roach, natproach@gmail.com
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
13
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
14 import sys
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
15 from math import isnan, log10
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
16
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
17 from scipy.stats import fisher_exact
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
18
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
19
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
20 def pval_to_phredqual(pval):
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
21 return int(round(-10. * log10(pval)))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
22
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
23
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
24 def parseSimpleSNPpileup(fields, ref_base, alt_base):
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
25 base_to_idx = {
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
26 'A': 0,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
27 'a': 0,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
28 'T': 1,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
29 't': 1,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
30 'C': 2,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
31 'c': 2,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
32 'G': 3,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
33 'g': 3
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
34 }
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
35
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
36 base_to_idx_stranded = {
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
37 'A': 0,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
38 'T': 1,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
39 'C': 2,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
40 'G': 3,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
41 'a': 4,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
42 't': 5,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
43 'c': 6,
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
44 'g': 7
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
45 }
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
46 ref_base2 = fields[2]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
47 counts = [0, 0, 0, 0]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
48 stranded_counts = [0, 0, 0, 0, 0, 0, 0, 0]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
49 ref_idx = base_to_idx[fields[2]]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
50 dp = int(fields[3])
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
51 carrot_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
52 ins_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
53 ins_str = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
54 ins_len = 0
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
55 insertion = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
56 del_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
57 del_str = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
58 del_len = 0
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
59 deletion = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
60 # dollar_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
61 for base in fields[4]:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
62 if carrot_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
63 carrot_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
64 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
65 if ins_len > 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
66 insertion += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
67 ins_len -= 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
68 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
69 if del_len > 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
70 deletion += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
71 del_len -= 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
72 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
73 if ins_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
74 if base.isdigit():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
75 ins_str += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
76 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
77 ins_len = int(ins_str) - 1
6
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
78 ins_str = ""
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
79 insertion = base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
80 ins_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
81 elif del_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
82 if base.isdigit():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
83 del_str += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
84 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
85 del_len = int(del_str) - 1
6
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
86 del_str = ""
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
87 deletion = base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
88 del_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
89 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
90 if base == '^':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
91 carrot_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
92 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
93 elif base == '$':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
94 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
95 elif base == '+':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
96 ins_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
97 elif base == '-':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
98 del_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
99 elif base == '.':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
100 counts[ref_idx] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
101 stranded_counts[base_to_idx_stranded[ref_base2]] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
102 elif base == ',':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
103 counts[ref_idx] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
104 stranded_counts[base_to_idx_stranded[ref_base2.lower()]] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
105 elif base == 'N' or base == 'n':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
106 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
107 elif base == '*':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
108 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
109 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
110 counts[base_to_idx[base]] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
111 stranded_counts[base_to_idx_stranded[base]] += 1
6
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
112 if sum(counts) == 0:
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
113 af = float("nan")
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
114 else:
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
115 af = float(counts[base_to_idx[alt_base]]) / float(sum(counts))
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
116 if float(sum(stranded_counts[0:4])) == 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
117 faf = float("nan")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
118 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
119 faf = float(stranded_counts[base_to_idx_stranded[alt_base]]) / float(sum(stranded_counts[0:4]))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
120 if float(sum(stranded_counts[4:])) == 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
121 raf = float("nan")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
122 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
123 raf = float(stranded_counts[base_to_idx_stranded[alt_base.lower()]]) / float(sum(stranded_counts[4:]))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
124 dp4 = [stranded_counts[base_to_idx_stranded[ref_base]],
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
125 stranded_counts[base_to_idx_stranded[ref_base.lower()]],
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
126 stranded_counts[base_to_idx_stranded[alt_base]],
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
127 stranded_counts[base_to_idx_stranded[alt_base.lower()]]]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
128 return (dp, af, faf, raf, dp4)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
129
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
130
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
131 def parseIndelPileup(fields, ref_base, alt_base):
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
132 counts = [0, 0, 0, 0, 0, 0, 0, 0, 0] # indel ref match, indel fwd ref match, indel rev ref match, indel alt match, indel fwd alt match, indel rev alt match, other, other fwd, other rev
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
133 ref_base2 = fields[2]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
134
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
135 carrot_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
136 ins_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
137 ins_str = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
138 ins_len = 0
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
139 del_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
140 del_str = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
141 del_len = 0
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
142 first_iter = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
143 forward_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
144 last_seq = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
145 last_seq_code = 'b'
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
146 for base in fields[4]:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
147 if ins_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
148 if base.isdigit():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
149 ins_str += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
150 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
151 ins_len = int(ins_str)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
152 ins_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
153 if del_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
154 if base.isdigit():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
155 del_str += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
156 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
157 del_len = int(del_str)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
158 del_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
159 if ins_len > 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
160 last_seq += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
161 last_seq_code = 'i'
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
162 ins_len -= 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
163 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
164 if del_len > 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
165 last_seq += base
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
166 last_seq_code = 'd'
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
167 del_len -= 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
168 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
169 if carrot_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
170 carrot_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
171 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
172 if base == '.' or base == ','\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
173 or base == 'A' or base == 'a'\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
174 or base == 'C' or base == 'c'\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
175 or base == 'G' or base == 'g'\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
176 or base == 'T' or base == 't'\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
177 or base == 'N' or base == 'n'\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
178 or base == '>' or base == '<'\
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
179 or base == '*' or base == '#':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
180 if first_iter:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
181 first_iter = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
182 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
183 if last_seq_code == 'i':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
184 if last_seq.upper() == alt_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
185 counts[3] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
186 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
187 counts[4] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
188 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
189 counts[5] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
190 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
191 counts[6] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
192 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
193 counts[7] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
194 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
195 counts[8] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
196 elif last_seq_code == 'd':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
197 if last_seq.upper() == ref_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
198 counts[3] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
199 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
200 counts[4] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
201 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
202 counts[5] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
203 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
204 counts[6] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
205 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
206 counts[7] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
207 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
208 counts[8] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
209 elif last_seq_code == 'b':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
210 if last_seq.upper() == ref_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
211 counts[0] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
212 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
213 counts[1] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
214 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
215 counts[2] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
216 elif last_seq.upper() == alt_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
217 counts[3] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
218 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
219 counts[4] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
220 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
221 counts[5] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
222 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
223 counts[6] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
224 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
225 counts[7] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
226 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
227 counts[8] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
228 if base == '.':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
229 last_seq = ref_base2
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
230 forward_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
231 last_seq_code = 'b'
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
232 elif base == ',':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
233 last_seq = ref_base2
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
234 forward_flag = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
235 last_seq_code = 'b'
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
236 elif base == '>' or base == '<' or base == '*' or base == '#':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
237 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
238 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
239 forward_flag = base.isupper()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
240 last_seq = base.upper()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
241 last_seq_code = 'b'
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
242 elif base == '+':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
243 ins_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
244 ins_str = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
245 elif base == '-':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
246 del_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
247 del_str = ""
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
248 elif base == '^':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
249 carrot_flag = True
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
250 elif base == '$':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
251 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
252 if first_iter:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
253 first_iter = False
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
254
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
255 if last_seq_code == 'i':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
256 if last_seq.upper() == alt_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
257 counts[3] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
258 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
259 counts[4] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
260 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
261 counts[5] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
262 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
263 counts[6] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
264 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
265 counts[7] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
266 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
267 counts[8] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
268 elif last_seq_code == 'd':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
269 if last_seq.upper() == ref_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
270 counts[3] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
271 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
272 counts[4] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
273 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
274 counts[5] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
275 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
276 counts[6] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
277 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
278 counts[7] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
279 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
280 counts[8] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
281 elif last_seq_code == 'b':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
282 if last_seq.upper() == ref_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
283 counts[0] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
284 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
285 counts[1] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
286 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
287 counts[2] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
288 elif last_seq.upper() == alt_base.upper():
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
289 counts[3] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
290 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
291 counts[4] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
292 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
293 counts[5] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
294 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
295 counts[6] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
296 if forward_flag:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
297 counts[7] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
298 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
299 counts[8] += 1
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
300 dp = int(fields[3])
6
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
301 if sum([counts[0], counts[3], counts[6]]) == 0:
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
302 af = float("nan")
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
303 else:
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
304 af = float(counts[3]) / float(sum([counts[0], counts[3], counts[6]]))
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
305 if sum([counts[1], counts[4], counts[7]]) == 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
306 faf = float("nan")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
307 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
308 faf = float(counts[4]) / float(sum([counts[1], counts[4], counts[7]]))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
309 if sum([counts[2], counts[5], counts[8]]) == 0:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
310 raf = float("nan")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
311 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
312 raf = float(counts[5]) / float(sum([counts[2], counts[5], counts[8]]))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
313 dp4 = [counts[1], counts[2], counts[4], counts[5]]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
314 return (dp, af, faf, raf, dp4)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
315
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
316
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
317 def annotateVCF(in_vcf_filepath, in_mpileup_filepath, out_vcf_filepath):
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
318 in_vcf = open(in_vcf_filepath, 'r')
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
319 in_mpileup = open(in_mpileup_filepath, 'r')
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
320 out_vcf = open(out_vcf_filepath, 'w')
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
321
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
322 # First pass parsing of input vcf, output headerlines + new headerlines, add VCF sites we care about to to_examine (limits memory usage for sites that don't need annotation)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
323 to_examine = {}
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
324 for line in in_vcf:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
325 if line[0:2] == "##":
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
326 out_vcf.write(line)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
327 elif line[0] == "#":
6
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
328 out_vcf.write("##annotateVCFVersion=0.2\n")
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
329 out_vcf.write("##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Raw Depth\">\n")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
330 out_vcf.write("##INFO=<ID=AF,Number=1,Type=Float,Description=\"Allele Frequency\">\n")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
331 out_vcf.write("##INFO=<ID=FAF,Number=1,Type=Float,Description=\"Forward Allele Frequency\">\n")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
332 out_vcf.write("##INFO=<ID=RAF,Number=1,Type=Float,Description=\"Reverse Allele Frequency\">\n")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
333 out_vcf.write("##INFO=<ID=SB,Number=1,Type=Integer,Description=\"Phred-scaled strand bias at this position\">\n")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
334 out_vcf.write("##INFO=<ID=DP4,Number=4,Type=Integer,Description=\"Counts for ref-forward bases, ref-reverse, alt-forward and alt-reverse bases\">\n")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
335 out_vcf.write(line)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
336 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
337 fields = line.strip().split()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
338 if fields[0] in to_examine:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
339 to_examine[fields[0]][int(fields[1])] = (fields[3], fields[4])
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
340 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
341 to_examine[fields[0]] = {int(fields[1]): (fields[3], fields[4])}
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
342 in_vcf.close()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
343 data = {}
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
344
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
345 # Populate data dictionary, which relates chromosome and position to the following:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
346 # depth of coverage
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
347 # allele frequency
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
348 # forward strand allele frequency
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
349 # reverse strand allele frequency
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
350 # dp4 - depth of coverage of ref allele fwd strand, DOC of ref allele rev strand, DOC of alt allele fwd strand, DOC of alt allele rev strand
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
351 for line in in_mpileup:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
352 fields = line.strip().split()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
353 if fields[0] not in to_examine:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
354 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
355 if int(fields[1]) not in to_examine[fields[0]]:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
356 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
357 (ref_base, alt_base) = to_examine[fields[0]][int(fields[1])]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
358 if len(ref_base.split(',')) > 1: # Can't handle multiple ref alleles
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
359 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
360 if len(alt_base.split(',')) > 1: # Can't handle multiple alt alleles
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
361 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
362 if len(ref_base) > 1 or len(alt_base) > 1:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
363 if len(ref_base) > 1 and len(alt_base) > 1: # Can't handle complex indels
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
364 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
365 data[(fields[0], int(fields[1]))] = parseIndelPileup(fields, ref_base, alt_base)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
366 if len(ref_base) == 1 and len(alt_base) == 1:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
367 data[(fields[0], int(fields[1]))] = parseSimpleSNPpileup(fields, ref_base, alt_base)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
368 in_mpileup.close()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
369 # Reopen vcf, this time, skip header, annotate all the sites for which there is an entry in data dictionary
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
370 # (Sites without entries have either multiple ref or alt bases, or have complex indels. Not supported (for now), and not reported as a result)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
371 in_vcf = open(in_vcf_filepath, 'r')
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
372 for line in in_vcf:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
373 if line[0] == '#':
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
374 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
375 fields = line.strip().split('\t')
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
376 if (fields[0], int(fields[1])) not in data:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
377 continue
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
378 (dp, af, faf, raf, dp4) = data[(fields[0], int(fields[1]))]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
379 dp2x2 = [[dp4[0], dp4[1]], [dp4[2], dp4[3]]]
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
380 _, p_val = fisher_exact(dp2x2)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
381 sb = pval_to_phredqual(p_val)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
382 if fields[7] == "":
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
383 info = []
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
384 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
385 info = fields[7].split(';')
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
386 info.append("DP=%d" % (dp))
6
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
387 if isnan(af):
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
388 info.append("AF=NaN")
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
389 else:
22f6a0e7424f "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit 13769e7d51b30a1d15eb62a9ba89ee2064f3ddc3"
iuc
parents: 4
diff changeset
390 info.append("AF=%.6f" % (af))
4
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
391 if isnan(faf):
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
392 info.append("FAF=NaN")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
393 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
394 info.append("FAF=%.6f" % (faf))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
395 if isnan(raf):
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
396 info.append("RAF=NaN")
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
397 else:
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
398 info.append("RAF=%.6f" % (raf))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
399 info.append("SB=%d" % (sb))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
400 info.append("DP4=%s" % (','.join([str(x) for x in dp4])))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
401 new_info = ';'.join(info)
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
402 fields[7] = new_info
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
403 out_vcf.write("%s\n" % ("\t".join(fields)))
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
404 in_vcf.close()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
405 out_vcf.close()
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
406
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
407
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
408 if __name__ == "__main__":
84dcccebad3d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/medaka commit e80b649094384fc6d7a8f917300db3550cc99a44"
iuc
parents:
diff changeset
409 annotateVCF(sys.argv[1], sys.argv[2], sys.argv[3])