annotate rename.py @ 1:0d13e4410c3d draft

planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
author yating-l
date Mon, 18 Jun 2018 18:36:00 -0400
parents 65c9ce351343
children 8fdd3e06e1ec
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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 from collections import OrderedDict
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
7
1
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
8 def rename(inputfile, outputfile, writer):
0
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
9 namemap = OrderedDict()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
10 with open(outputfile, 'w') as out:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
11 with open(inputfile, 'r') as rf:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
12 lines = rf.readlines()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
13 i = 1
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
14 for line in lines:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
15 if ">" in line:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
16 oldname = line[1:].rstrip()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
17 newname = "scaffold_" + str(i)
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
18 line = ">" + newname
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
19 i = i+1
1
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
20 writer.writerow([oldname, newname])
0
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
21 #TODO: Add line breaks to chromosome sequences that are in a single line
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
22 out.write(line.rstrip() + "\n")
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
23
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
24 def main():
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
25 inputfile = str(sys.argv[1])
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
26 outputfile = str(sys.argv[2])
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
27 indexfile = str(sys.argv[3])
1
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
28 csvfile = open(indexfile, 'w')
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
29 fieldnames = ['Original sequence name', 'Renamed sequence name']
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
30 writer = csv.writer(csvfile)
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
31 writer.writerow(fieldnames)
0d13e4410c3d planemo upload commit 68b7e251486853b77396932b588c20f389c366d2
yating-l
parents: 0
diff changeset
32 rename(inputfile, outputfile, writer)
0
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
33
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
34 if __name__ == "__main__":
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
35 main()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
36
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
37