Mercurial > repos > yating-l > rename_scaffolds
annotate rename.py @ 2:8fdd3e06e1ec draft
planemo upload commit eb7c6866b218a41d9b22255ee6afb83551a5ff40
author | yating-l |
---|---|
date | Mon, 25 Jun 2018 15:29:18 -0400 |
parents | 0d13e4410c3d |
children | 9529a207d704 |
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 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
3 """ |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
4 import sys |
1
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
5 import csv |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
6 |
1
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
7 def rename(inputfile, outputfile, writer): |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
8 with open(outputfile, 'w') as out: |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
9 with open(inputfile, 'r') as rf: |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
10 lines = rf.readlines() |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
11 i = 1 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
12 for line in lines: |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
13 if ">" in line: |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
14 oldname = line[1:].rstrip() |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
15 newname = "scaffold_" + str(i) |
2
8fdd3e06e1ec
planemo upload commit eb7c6866b218a41d9b22255ee6afb83551a5ff40
yating-l
parents:
1
diff
changeset
|
16 line = ">" + newname + "\n" |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
17 i = i+1 |
1
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
18 writer.writerow([oldname, newname]) |
2
8fdd3e06e1ec
planemo upload commit eb7c6866b218a41d9b22255ee6afb83551a5ff40
yating-l
parents:
1
diff
changeset
|
19 out.write(line) |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
20 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
21 def main(): |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
22 inputfile = str(sys.argv[1]) |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
23 outputfile = str(sys.argv[2]) |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
24 indexfile = str(sys.argv[3]) |
1
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
25 csvfile = open(indexfile, 'w') |
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
26 writer = csv.writer(csvfile) |
0d13e4410c3d
planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents:
0
diff
changeset
|
27 rename(inputfile, outputfile, writer) |
0
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
28 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
29 if __name__ == "__main__": |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
30 main() |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
31 |
65c9ce351343
planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff
changeset
|
32 |