Mercurial > repos > yating-l > rename_scaffolds
annotate rename.py @ 1:148e353417e7 draft default tip
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
author | yating-l |
---|---|
date | Fri, 30 Dec 2016 17:50:54 -0500 |
parents | 1a0e39acb62f |
children |
rev | line source |
---|---|
0
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
1 """ |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
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 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
3 """ |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
4 import sys |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
5 from collections import OrderedDict |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
6 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
7 def rename(inputfile, outputfile, indexfile): |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
8 namemap = OrderedDict() |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
9 with open(outputfile, 'w') as out: |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
10 with open(inputfile, 'r') as rf: |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
11 lines = rf.readlines() |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
12 i = 1 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
13 for line in lines: |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
14 if ">" in line: |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
15 oldname = line[1:].rstrip() |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
16 newname = "scaffold_" + str(i) |
1
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
17 line = ">" + newname |
0
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
18 i = i+1 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
19 namemap[oldname] = newname |
1
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
20 #TODO: Add line breaks to chromosome sequences that are in a single line |
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
21 #else: |
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
22 #if (len(line) > 50): |
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
23 #for |
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
24 out.write(line.rstrip() + "\n") |
0
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
25 with open(indexfile, 'w') as index: |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
26 for k in namemap: |
1
148e353417e7
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
0
diff
changeset
|
27 index.write(k + "\t" + namemap[k] + "\n") |
0
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
28 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
29 def main(): |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
30 inputfile = str(sys.argv[1]) |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
31 outputfile = str(sys.argv[2]) |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
32 indexfile = str(sys.argv[3]) |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
33 rename(inputfile, outputfile, indexfile) |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
34 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
35 if __name__ == "__main__": |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
36 main() |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
37 |
1a0e39acb62f
planemo upload commit 3431b38459716fd36c193c1035ede040f7ca2e87-dirty
yating-l
parents:
diff
changeset
|
38 |