Mercurial > repos > yating-l > rename_scaffolds
annotate rename.py @ 4:e35a3509c160 draft
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
author | yating-l |
---|---|
date | Thu, 26 Jul 2018 15:46:24 -0400 |
parents | 9529a207d704 |
children | 7c8b327f298c |
rev | line source |
---|---|
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
1 """ |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
2 Call rename to rename scaffolds in reference genome so that the sequence names are less than 31 characters. Rename all scaffolds to scaffold_1, scaffold_2, ..., scaffold_N and provide a name mapping file |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
3 Call truncate to truncate the scaffold names that are more than 31 characters. Replace each invalid character (non-ASCII, '\t', '\n', '\x0b', '\x0c', '\r') with '_' |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
4 """ |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
5 import sys |
1
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
6 import csv |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
7 import codecs |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
8 import string |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
9 |
1
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
10 def rename(inputfile, outputfile, writer): |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
11 with open(outputfile, 'w') as out: |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
12 with codecs.open(inputfile, 'r', encoding='utf-8') as rf: |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
13 lines = rf.readlines() |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
14 i = 1 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
15 for line in lines: |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
16 if ">" in line: |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
17 oldname = line[1:].rstrip() |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
18 newname = "scaffold_" + str(i) |
2
8fdd3e06e1ec
planemo upload commit eb7c6866b218a41d9b22255ee6afb83551a5ff40
yating-l
parents:
1
diff
changeset
|
19 line = ">" + newname + "\n" |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
20 i = i+1 |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
21 writer.writerow([oldname.encode('utf-8'), newname]) |
2
8fdd3e06e1ec
planemo upload commit eb7c6866b218a41d9b22255ee6afb83551a5ff40
yating-l
parents:
1
diff
changeset
|
22 out.write(line) |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
23 |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
24 def truncate(inputFile, outputFile, valid_characters): |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
25 names = [] |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
26 with codecs.open(outputFile, 'w', encoding='utf-8') as out: |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
27 with codecs.open(inputFile, 'r', encoding='utf-8') as rf: |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
28 lines = rf.readlines() |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
29 for l in lines: |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
30 if ">" in l: |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
31 print l.encode('utf-8') |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
32 name = l[1:].rstrip() |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
33 name = substituteNonAscii(name, valid_characters) |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
34 if len(name) > 31: |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
35 name = name[:31] |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
36 print "\tTruncate the scaffold name to less than 31 characters: %s" % name |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
37 if name in names: |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
38 sys.exit("Name conflict! Name " + name + " already exist.") |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
39 names.append(name) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
40 l = ">" + name + "\n" |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
41 print "======================\n" |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
42 out.write(l) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
43 |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
44 def substituteNonAscii(str, valid_charaters): |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
45 l = [] |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
46 for c in str: |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
47 if c not in valid_charaters: |
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
48 print "\tSubstitute invalid character %s with _" % c.encode('utf-8') |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
49 c = '_' |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
50 l.append(c) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
51 return "".join(l) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
52 |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
53 def main(): |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
54 inputfile = str(sys.argv[1]) |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
55 manipulate = str(sys.argv[2]) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
56 outputfile = str(sys.argv[3]) |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
57 valid_characters = string.letters + string.punctuation + string.digits + ' ' |
3
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
58 if manipulate == "rename": |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
59 indexfile = str(sys.argv[4]) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
60 csvfile = open(indexfile, 'w') |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
61 writer = csv.writer(csvfile) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
62 rename(inputfile, outputfile, writer) |
9529a207d704
planemo upload commit fb7ce0e9f333b753a39ff171a8fbf39c3b3a069c
yating-l
parents:
2
diff
changeset
|
63 elif manipulate == "truncate": |
4
e35a3509c160
planemo upload commit 3675b4447d11fb1cd75d505886e1bf693f9d07f5
yating-l
parents:
3
diff
changeset
|
64 truncate(inputfile, outputfile, valid_characters) |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
65 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
66 if __name__ == "__main__": |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
67 main() |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
68 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
69 |