Mercurial > repos > rhpvorderman > shm_csr

annotate shm_csr.py @ 5:495a521cf9f2 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 40d62ec3d2fea3c3bac605c976941e1f3b7e2cd4"
author rhpvorderman
date Tue, 16 Nov 2021 11:16:06 +0000
parents 64d74ba01a7c
children ea9d5fc4c001
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
1 import argparse
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
2 import logging
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
3 import sys
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
4 import os
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
5 import re
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
6
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
7 from collections import defaultdict
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
8
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
9 def main():
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
10 parser = argparse.ArgumentParser()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
11 parser.add_argument("--input", help="The '7_V-REGION-mutation-and-AA-change-table' and '10_V-REGION-mutation-hotspots' merged together, with an added 'best_match' annotation")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
12 parser.add_argument("--genes", help="The genes available in the 'best_match' column")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
13 parser.add_argument("--empty_region_filter", help="Where does the sequence start?", choices=['leader', 'FR1', 'CDR1', 'FR2'])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
14 parser.add_argument("--output", help="Output file")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
15
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
16 args = parser.parse_args()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
17
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
18 infile = args.input
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
19 genes = str(args.genes).split(",")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
20 empty_region_filter = args.empty_region_filter
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
21 outfile = args.output
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
22
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
23 genedic = dict()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
24
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
25 mutationdic = dict()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
26 mutationMatcher = re.compile("^(.)(\d+).(.),?[ ]?(.)?(\d+)?.?(.)?(.?.?.?.?.?)?")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
27 mutationMatcher = re.compile("^([actg])(\d+).([actg]),?[ ]?([A-Z])?(\d+)?.?([A-Z])?(.*)?")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
28 mutationMatcher = re.compile("^([actg])(\d+).([actg]),?[ ]?([A-Z])?(\d+)?[>]?([A-Z;])?(.*)?")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
29 mutationMatcher = re.compile(r"^([nactg])(\d+).([nactg]),?[ ]?([A-Z*])?(\d+)?[>]?([A-Z*;])?(.*)?")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
30 NAMatchResult = (None, None, None, None, None, None, '')
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
31 geneMatchers = {gene: re.compile("^" + gene + ".*") for gene in genes}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
32 linecount = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
33
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
34 IDIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
35 best_matchIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
36 fr1Index = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
37 cdr1Index = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
38 fr2Index = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
39 cdr2Index = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
40 fr3Index = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
41 first = True
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
42 IDlist = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
43 mutationList = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
44 mutationListByID = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
45 cdr1LengthDic = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
46 cdr2LengthDic = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
47
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
48 fr1LengthDict = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
49 fr2LengthDict = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
50 fr3LengthDict = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
51
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
52 cdr1LengthIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
53 cdr2LengthIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
54
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
55 fr1SeqIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
56 fr2SeqIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
57 fr3SeqIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
58
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
59 tandem_sum_by_class = defaultdict(int)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
60 expected_tandem_sum_by_class = defaultdict(float)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
61
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
62 with open(infile, 'r') as i:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
63 for line in i:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
64 if first:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
65 linesplt = line.split("\t")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
66 IDIndex = linesplt.index("Sequence.ID")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
67 best_matchIndex = linesplt.index("best_match")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
68 fr1Index = linesplt.index("FR1.IMGT")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
69 cdr1Index = linesplt.index("CDR1.IMGT")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
70 fr2Index = linesplt.index("FR2.IMGT")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
71 cdr2Index = linesplt.index("CDR2.IMGT")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
72 fr3Index = linesplt.index("FR3.IMGT")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
73 cdr1LengthIndex = linesplt.index("CDR1.IMGT.length")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
74 cdr2LengthIndex = linesplt.index("CDR2.IMGT.length")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
75 fr1SeqIndex = linesplt.index("FR1.IMGT.seq")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
76 fr2SeqIndex = linesplt.index("FR2.IMGT.seq")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
77 fr3SeqIndex = linesplt.index("FR3.IMGT.seq")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
78 first = False
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
79 continue
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
80 linecount += 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
81 linesplt = line.split("\t")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
82 ID = linesplt[IDIndex]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
83 genedic[ID] = linesplt[best_matchIndex]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
84
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
85 mutationdic[ID + "_FR1"] = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
86 if len(linesplt[fr1Index]) > 5 and empty_region_filter == "leader":
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
87 mutationdic[ID + "_FR1"] = [mutationMatcher.match(x).groups() for x in linesplt[fr1Index].split("|") if x]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
88
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
89 mutationdic[ID + "_CDR1"] = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
90 if len(linesplt[cdr1Index]) > 5 and empty_region_filter in ["leader", "FR1"]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
91 mutationdic[ID + "_CDR1"] = [mutationMatcher.match(x).groups() for x in linesplt[cdr1Index].split("|") if x]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
92
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
93 mutationdic[ID + "_FR2"] = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
94 if len(linesplt[fr2Index]) > 5 and empty_region_filter in ["leader", "FR1", "CDR1"]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
95 mutationdic[ID + "_FR2"] = [mutationMatcher.match(x).groups() for x in linesplt[fr2Index].split("|") if x]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
96
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
97 mutationdic[ID + "_CDR2"] = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
98 if len(linesplt[cdr2Index]) > 5:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
99 mutationdic[ID + "_CDR2"] = [mutationMatcher.match(x).groups() for x in linesplt[cdr2Index].split("|") if x]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
100
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
101 mutationdic[ID + "_FR2-CDR2"] = mutationdic[ID + "_FR2"] + mutationdic[ID + "_CDR2"]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
102
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
103 mutationdic[ID + "_FR3"] = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
104 if len(linesplt[fr3Index]) > 5:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
105 mutationdic[ID + "_FR3"] = [mutationMatcher.match(x).groups() for x in linesplt[fr3Index].split("|") if x]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
106
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
107 mutationList += mutationdic[ID + "_FR1"] + mutationdic[ID + "_CDR1"] + mutationdic[ID + "_FR2"] + mutationdic[ID + "_CDR2"] + mutationdic[ID + "_FR3"]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
108 mutationListByID[ID] = mutationdic[ID + "_FR1"] + mutationdic[ID + "_CDR1"] + mutationdic[ID + "_FR2"] + mutationdic[ID + "_CDR2"] + mutationdic[ID + "_FR3"]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
109
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
110 try:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
111 cdr1Length = int(linesplt[cdr1LengthIndex])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
112 except:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
113 cdr1Length = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
114
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
115 try:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
116 cdr2Length = int(linesplt[cdr2LengthIndex])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
117 except:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
118 cdr2Length = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
119
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
120 #print linesplt[fr2SeqIndex]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
121 fr1Length = len(linesplt[fr1SeqIndex]) if empty_region_filter == "leader" else 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
122 fr2Length = len(linesplt[fr2SeqIndex]) if empty_region_filter in ["leader", "FR1", "CDR1"] else 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
123 fr3Length = len(linesplt[fr3SeqIndex])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
124
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
125 cdr1LengthDic[ID] = cdr1Length
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
126 cdr2LengthDic[ID] = cdr2Length
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
127
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
128 fr1LengthDict[ID] = fr1Length
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
129 fr2LengthDict[ID] = fr2Length
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
130 fr3LengthDict[ID] = fr3Length
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
131
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
132 IDlist += [ID]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
133 print("len(mutationdic) =", len(mutationdic))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
134
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
135 with open(os.path.join(os.path.dirname(os.path.abspath(infile)), "mutationdict.txt"), 'w') as out_handle:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
136 for ID, lst in mutationdic.items():
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
137 for mut in lst:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
138 out_handle.write("{0}\t{1}\n".format(ID, "\t".join([str(x) for x in mut])))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
139
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
140 #tandem mutation stuff
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
141 tandem_frequency = defaultdict(int)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
142 mutation_frequency = defaultdict(int)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
143
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
144 mutations_by_id_dic = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
145 first = True
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
146 mutation_by_id_file = os.path.join(os.path.dirname(outfile), "mutation_by_id.txt")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
147 with open(mutation_by_id_file, 'r') as mutation_by_id:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
148 for l in mutation_by_id:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
149 if first:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
150 first = False
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
151 continue
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
152 splt = l.split("\t")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
153 mutations_by_id_dic[splt[0]] = int(splt[1])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
154
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
155 tandem_file = os.path.join(os.path.dirname(outfile), "tandems_by_id.txt")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
156 with open(tandem_file, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
157 highest_tandem_length = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
158
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
159 o.write("Sequence.ID\tnumber_of_mutations\tnumber_of_tandems\tregion_length\texpected_tandems\tlongest_tandem\ttandems\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
160 for ID in IDlist:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
161 mutations = mutationListByID[ID]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
162 if len(mutations) == 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
163 continue
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
164 last_mut = max(mutations, key=lambda x: int(x[1]))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
165
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
166 last_mut_pos = int(last_mut[1])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
167
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
168 mut_positions = [False] * (last_mut_pos + 1)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
169
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
170 for mutation in mutations:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
171 frm, where, to, frmAA, whereAA, toAA, thing = mutation
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
172 where = int(where)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
173 mut_positions[where] = True
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
174
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
175 tandem_muts = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
176 tandem_start = -1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
177 tandem_length = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
178 for i in range(len(mut_positions)):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
179 if mut_positions[i]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
180 if tandem_start == -1:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
181 tandem_start = i
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
182 tandem_length += 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
183 #print "".join(["1" if x else "0" for x in mut_positions[:i+1]])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
184 else:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
185 if tandem_length > 1:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
186 tandem_muts.append((tandem_start, tandem_length))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
187 #print "{0}{1} {2}:{3}".format(" " * (i - tandem_length), "^" * tandem_length, tandem_start, tandem_length)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
188 tandem_start = -1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
189 tandem_length = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
190 if tandem_length > 1: # if the sequence ends with a tandem mutation
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
191 tandem_muts.append((tandem_start, tandem_length))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
192
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
193 if len(tandem_muts) > 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
194 if highest_tandem_length < len(tandem_muts):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
195 highest_tandem_length = len(tandem_muts)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
196
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
197 region_length = fr1LengthDict[ID] + cdr1LengthDic[ID] + fr2LengthDict[ID] + cdr2LengthDic[ID] + fr3LengthDict[ID]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
198 longest_tandem = max(tandem_muts, key=lambda x: x[1]) if len(tandem_muts) else (0, 0)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
199 num_mutations = mutations_by_id_dic[ID] # len(mutations)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
200 f_num_mutations = float(num_mutations)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
201 num_tandem_muts = len(tandem_muts)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
202 expected_tandem_muts = f_num_mutations * (f_num_mutations - 1.0) / float(region_length)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
203 o.write("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\n".format(ID,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
204 str(num_mutations),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
205 str(num_tandem_muts),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
206 str(region_length),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
207 str(round(expected_tandem_muts, 2)),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
208 str(longest_tandem[1]),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
209 str(tandem_muts)))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
210 gene = genedic[ID]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
211 if gene.find("unmatched") == -1:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
212 tandem_sum_by_class[gene] += num_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
213 expected_tandem_sum_by_class[gene] += expected_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
214
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
215 tandem_sum_by_class["all"] += num_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
216 expected_tandem_sum_by_class["all"] += expected_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
217
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
218 gene = gene[:3]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
219 if gene in ["IGA", "IGG"]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
220 tandem_sum_by_class[gene] += num_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
221 expected_tandem_sum_by_class[gene] += expected_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
222 else:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
223 tandem_sum_by_class["unmatched"] += num_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
224 expected_tandem_sum_by_class["unmatched"] += expected_tandem_muts
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
225
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
226
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
227 for tandem_mut in tandem_muts:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
228 tandem_frequency[str(tandem_mut[1])] += 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
229 #print "\t".join([ID, str(len(tandem_muts)), str(longest_tandem[1]) , str(tandem_muts)])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
230
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
231 tandem_freq_file = os.path.join(os.path.dirname(outfile), "tandem_frequency.txt")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
232 with open(tandem_freq_file, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
233 for frq in sorted([int(x) for x in list(tandem_frequency.keys())]):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
234 o.write("{0}\t{1}\n".format(frq, tandem_frequency[str(frq)]))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
235
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
236 tandem_row = []
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
237 genes_extra = list(genes)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
238 genes_extra.append("all")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
239 for x, y, in zip([tandem_sum_by_class[x] for x in genes_extra], [expected_tandem_sum_by_class[x] for x in genes_extra]):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
240 if y != 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
241 tandem_row += [x, round(y, 2), round(x / y, 2)]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
242 else:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
243 tandem_row += [x, round(y, 2), 0]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
244
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
245 tandem_freq_file = os.path.join(os.path.dirname(outfile), "shm_overview_tandem_row.txt")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
246 with open(tandem_freq_file, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
247 o.write("Tandems/Expected (ratio),{0}\n".format(",".join([str(x) for x in tandem_row])))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
248
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
249 #print mutationList, linecount
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
250
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
251 AALength = (int(max(mutationList, key=lambda i: int(i[4]) if i[4] and i[5] != ";" else 0)[4]) + 1) # [4] is the position of the AA mutation, None if silent
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
252 if AALength < 60:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
253 AALength = 64
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
254
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
255 AA_mutation = [0] * AALength
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
256 AA_mutation_dic = {"IGA": AA_mutation[:], "IGG": AA_mutation[:], "IGM": AA_mutation[:], "IGE": AA_mutation[:], "unm": AA_mutation[:], "all": AA_mutation[:]}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
257 AA_mutation_empty = AA_mutation[:]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
258
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
259 print("AALength:", AALength)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
260 aa_mutations_by_id_file = outfile[:outfile.rindex("/")] + "/aa_id_mutations.txt"
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
261 with open(aa_mutations_by_id_file, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
262 o.write("ID\tbest_match\t" + "\t".join([str(x) for x in range(1,AALength)]) + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
263 for ID in list(mutationListByID.keys()):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
264 AA_mutation_for_ID = AA_mutation_empty[:]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
265 for mutation in mutationListByID[ID]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
266 if mutation[4] and mutation[5] != ";":
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
267 AA_mutation_position = int(mutation[4])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
268 try:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
269 AA_mutation[AA_mutation_position] += 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
270 AA_mutation_for_ID[AA_mutation_position] += 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
271 except Exception as e:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
272 print(e)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
273 print(mutation)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
274 sys.exit()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
275 clss = genedic[ID][:3]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
276 AA_mutation_dic[clss][AA_mutation_position] += 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
277 o.write(ID + "\t" + genedic[ID] + "\t" + "\t".join([str(x) for x in AA_mutation_for_ID[1:]]) + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
278
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
279
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
280
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
281 #absent AA stuff
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
282 absentAACDR1Dic = defaultdict(list)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
283 absentAACDR1Dic[5] = list(range(29,36))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
284 absentAACDR1Dic[6] = list(range(29,35))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
285 absentAACDR1Dic[7] = list(range(30,35))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
286 absentAACDR1Dic[8] = list(range(30,34))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
287 absentAACDR1Dic[9] = list(range(31,34))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
288 absentAACDR1Dic[10] = list(range(31,33))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
289 absentAACDR1Dic[11] = [32]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
290
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
291 absentAACDR2Dic = defaultdict(list)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
292 absentAACDR2Dic[0] = list(range(55,65))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
293 absentAACDR2Dic[1] = list(range(56,65))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
294 absentAACDR2Dic[2] = list(range(56,64))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
295 absentAACDR2Dic[3] = list(range(57,64))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
296 absentAACDR2Dic[4] = list(range(57,63))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
297 absentAACDR2Dic[5] = list(range(58,63))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
298 absentAACDR2Dic[6] = list(range(58,62))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
299 absentAACDR2Dic[7] = list(range(59,62))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
300 absentAACDR2Dic[8] = list(range(59,61))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
301 absentAACDR2Dic[9] = [60]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
302
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
303 absentAA = [len(IDlist)] * (AALength-1)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
304 for k, cdr1Length in cdr1LengthDic.items():
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
305 for c in absentAACDR1Dic[cdr1Length]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
306 absentAA[c] -= 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
307
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
308 for k, cdr2Length in cdr2LengthDic.items():
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
309 for c in absentAACDR2Dic[cdr2Length]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
310 absentAA[c] -= 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
311
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
312
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
313 aa_mutations_by_id_file = outfile[:outfile.rindex("/")] + "/absent_aa_id.txt"
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
314 with open(aa_mutations_by_id_file, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
315 o.write("ID\tcdr1length\tcdr2length\tbest_match\t" + "\t".join([str(x) for x in range(1,AALength)]) + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
316 for ID in IDlist:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
317 absentAAbyID = [1] * (AALength-1)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
318 cdr1Length = cdr1LengthDic[ID]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
319 for c in absentAACDR1Dic[cdr1Length]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
320 absentAAbyID[c] -= 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
321
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
322 cdr2Length = cdr2LengthDic[ID]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
323 for c in absentAACDR2Dic[cdr2Length]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
324 absentAAbyID[c] -= 1
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
325 o.write(ID + "\t" + str(cdr1Length) + "\t" + str(cdr2Length) + "\t" + genedic[ID] + "\t" + "\t".join([str(x) for x in absentAAbyID]) + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
326
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
327 if linecount == 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
328 print("No data, exiting")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
329 with open(outfile, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
330 o.write("RGYW (%)," + ("0,0,0\n" * len(genes)))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
331 o.write("WRCY (%)," + ("0,0,0\n" * len(genes)))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
332 o.write("WA (%)," + ("0,0,0\n" * len(genes)))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
333 o.write("TW (%)," + ("0,0,0\n" * len(genes)))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
334 sys.exit()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
335
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
336 hotspotMatcher = re.compile("[actg]+,(\d+)-(\d+)\((.*)\)")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
337 RGYWCount = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
338 WRCYCount = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
339 WACount = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
340 TWCount = {}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
341
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
342 #IDIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
343 ataIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
344 tatIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
345 aggctatIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
346 atagcctIndex = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
347 first = True
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
348 with open(infile, 'r') as i:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
349 for line in i:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
350 if first:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
351 linesplt = line.split("\t")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
352 ataIndex = linesplt.index("X.a.t.a")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
353 tatIndex = linesplt.index("t.a.t.")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
354 aggctatIndex = linesplt.index("X.a.g.g.c.t..a.t.")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
355 atagcctIndex = linesplt.index("X.a.t..a.g.c.c.t.")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
356 first = False
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
357 continue
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
358 linesplt = line.split("\t")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
359 gene = linesplt[best_matchIndex]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
360 ID = linesplt[IDIndex]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
361 RGYW = [(int(x), int(y), z) for (x, y, z) in
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
362 [hotspotMatcher.match(x).groups() for x in linesplt[aggctatIndex].split("|") if x]]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
363 WRCY = [(int(x), int(y), z) for (x, y, z) in
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
364 [hotspotMatcher.match(x).groups() for x in linesplt[atagcctIndex].split("|") if x]]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
365 WA = [(int(x), int(y), z) for (x, y, z) in
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
366 [hotspotMatcher.match(x).groups() for x in linesplt[ataIndex].split("|") if x]]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
367 TW = [(int(x), int(y), z) for (x, y, z) in
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
368 [hotspotMatcher.match(x).groups() for x in linesplt[tatIndex].split("|") if x]]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
369 RGYWCount[ID], WRCYCount[ID], WACount[ID], TWCount[ID] = 0, 0, 0, 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
370
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
371 with open(os.path.join(os.path.dirname(os.path.abspath(infile)), "RGYW.txt"), 'a') as out_handle:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
372 for hotspot in RGYW:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
373 out_handle.write("{0}\t{1}\n".format(ID, "\t".join([str(x) for x in hotspot])))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
374
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
375 mutationList = mutationdic[ID + "_FR1"] + mutationdic[ID + "_CDR1"] + mutationdic[ID + "_FR2"] + mutationdic[ID + "_CDR2"] + mutationdic[ID + "_FR3"]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
376 for mutation in mutationList:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
377 frm, where, to, AAfrm, AAwhere, AAto, junk = mutation
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
378 mutation_in_RGYW = any(((start <= int(where) <= end) for (start, end, region) in RGYW))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
379 mutation_in_WRCY = any(((start <= int(where) <= end) for (start, end, region) in WRCY))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
380 mutation_in_WA = any(((start <= int(where) <= end) for (start, end, region) in WA))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
381 mutation_in_TW = any(((start <= int(where) <= end) for (start, end, region) in TW))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
382
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
383 in_how_many_motifs = sum([mutation_in_RGYW, mutation_in_WRCY, mutation_in_WA, mutation_in_TW])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
384
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
385 if in_how_many_motifs > 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
386 RGYWCount[ID] += (1.0 * int(mutation_in_RGYW)) / in_how_many_motifs
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
387 WRCYCount[ID] += (1.0 * int(mutation_in_WRCY)) / in_how_many_motifs
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
388 WACount[ID] += (1.0 * int(mutation_in_WA)) / in_how_many_motifs
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
389 TWCount[ID] += (1.0 * int(mutation_in_TW)) / in_how_many_motifs
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
390
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
391 mutations_in_motifs_file = os.path.join(os.path.dirname(os.path.abspath(infile)), "mutation_in_motifs.txt")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
392 if not os.path.exists(mutation_by_id_file):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
393 with open(mutations_in_motifs_file, 'w') as out_handle:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
394 out_handle.write("{0}\n".format("\t".join([
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
395 "Sequence.ID",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
396 "mutation_position",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
397 "region",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
398 "from_nt",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
399 "to_nt",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
400 "mutation_position_AA",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
401 "from_AA",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
402 "to_AA",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
403 "motif",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
404 "motif_start_nt",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
405 "motif_end_nt",
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
406 "rest"
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
407 ])))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
408
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
409 with open(mutations_in_motifs_file, 'a') as out_handle:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
410 motif_dic = {"RGYW": RGYW, "WRCY": WRCY, "WA": WA, "TW": TW}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
411 for mutation in mutationList:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
412 frm, where, to, AAfrm, AAwhere, AAto, junk = mutation
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
413 for motif in list(motif_dic.keys()):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
414
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
415 for start, end, region in motif_dic[motif]:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
416 if start <= int(where) <= end:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
417 out_handle.write("{0}\n".format(
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
418 "\t".join([
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
419 ID,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
420 where,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
421 region,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
422 frm,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
423 to,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
424 str(AAwhere),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
425 str(AAfrm),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
426 str(AAto),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
427 motif,
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
428 str(start),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
429 str(end),
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
430 str(junk)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
431 ])
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
432 ))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
433
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
434
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
435
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
436 def mean(lst):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
437 return (float(sum(lst)) / len(lst)) if len(lst) > 0 else 0.0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
438
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
439
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
440 def median(lst):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
441 lst = sorted(lst)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
442 l = len(lst)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
443 if l == 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
444 return 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
445 if l == 1:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
446 return lst[0]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
447
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
448 l = int(l / 2)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
449
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
450 if len(lst) % 2 == 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
451 return float(lst[l] + lst[(l - 1)]) / 2.0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
452 else:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
453 return lst[l]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
454
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
455 funcs = {"mean": mean, "median": median, "sum": sum}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
456
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
457 directory = outfile[:outfile.rfind("/") + 1]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
458 value = 0
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
459 valuedic = dict()
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
460
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
461 for fname in list(funcs.keys()):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
462 for gene in genes:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
463 with open(directory + gene + "_" + fname + "_value.txt", 'r') as v:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
464 valuedic[gene + "_" + fname] = float(v.readlines()[0].rstrip())
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
465 with open(directory + "all_" + fname + "_value.txt", 'r') as v:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
466 valuedic["total_" + fname] = float(v.readlines()[0].rstrip())
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
467
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
468
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
469 def get_xyz(lst, gene, f, fname):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
470 x = round(round(f(lst), 1))
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
471 y = valuedic[gene + "_" + fname]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
472 z = str(round(x / float(y) * 100, 1)) if y != 0 else "0"
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
473 return (str(x), str(y), z)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
474
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
475 dic = {"RGYW": RGYWCount, "WRCY": WRCYCount, "WA": WACount, "TW": TWCount}
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
476 arr = ["RGYW", "WRCY", "WA", "TW"]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
477
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
478 for fname in list(funcs.keys()):
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
479 func = funcs[fname]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
480 foutfile = outfile[:outfile.rindex("/")] + "/hotspot_analysis_" + fname + ".txt"
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
481 with open(foutfile, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
482 for typ in arr:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
483 o.write(typ + " (%)")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
484 curr = dic[typ]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
485 for gene in genes:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
486 geneMatcher = geneMatchers[gene]
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
487 if valuedic[gene + "_" + fname] is 0:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
488 o.write(",0,0,0")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
489 else:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
490 x, y, z = get_xyz([curr[x] for x in [y for y, z in genedic.items() if geneMatcher.match(z)]], gene, func, fname)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
491 o.write("," + x + "," + y + "," + z)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
492 x, y, z = get_xyz([y for x, y in curr.items() if not genedic[x].startswith("unmatched")], "total", func, fname)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
493 #x, y, z = get_xyz([y for x, y in curr.iteritems()], "total", func, fname)
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
494 o.write("," + x + "," + y + "," + z + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
495
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
496
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
497 # for testing
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
498 seq_motif_file = outfile[:outfile.rindex("/")] + "/motif_per_seq.txt"
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
499 with open(seq_motif_file, 'w') as o:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
500 o.write("ID\tRGYW\tWRCY\tWA\tTW\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
501 for ID in IDlist:
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
502 #o.write(ID + "\t" + str(round(RGYWCount[ID], 2)) + "\t" + str(round(WRCYCount[ID], 2)) + "\t" + str(round(WACount[ID], 2)) + "\t" + str(round(TWCount[ID], 2)) + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
503 o.write(ID + "\t" + str(RGYWCount[ID]) + "\t" + str(WRCYCount[ID]) + "\t" + str(WACount[ID]) + "\t" + str(TWCount[ID]) + "\n")
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
504
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
505 if __name__ == "__main__":
64d74ba01a7c "planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff changeset
506 main()