Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics
annotate orthologs/ucsb_hamster/unbuild.py @ 0:5b9a38ec4a39 draft default tip
First commit of old repositories
author | osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> |
---|---|
date | Tue, 11 Mar 2014 12:19:13 -0700 |
parents | |
children |
rev | line source |
---|---|
0
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
1 #!/usr/bin/env python |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
2 ''' |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
3 Usage: unbuild.py outputDirectoryName hmmFileWithTabs |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
4 This script takes a tab sperated, newline escaped HMM models file then splits the models up and makes a file list. |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
5 The output files are put into the directory outputDirectoryName. |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
6 ''' |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
7 from sys import argv, exit |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
8 import os |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
9 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
10 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
11 def parse(line): |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
12 line = line.partition('\t') |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
13 name = line[0] + ".hmm" |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
14 data = line[2].replace("\\n", "\n") |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
15 return name, data |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
16 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
17 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
18 def writeFile(location, data): |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
19 try: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
20 with open(location, 'w') as f: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
21 f.write(data) |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
22 except: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
23 pass |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
24 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
25 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
26 def createDir(dir): |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
27 if not os.path.exists(dir): |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
28 os.makedirs(dir) |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
29 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
30 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
31 if __name__ == "__main__": |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
32 if len(argv) != 3: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
33 print "Usage: unbuild.py outputDirectoryName hmmFileWithTabs" |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
34 exit() |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
35 dir = argv[1] |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
36 createDir(dir) |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
37 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
38 name_list = [] # Stores new file names |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
39 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
40 # Read data and seperate into files. |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
41 with open(argv[2], 'r') as f: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
42 for line in f: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
43 name, data = parse(line) |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
44 if data.strip() == "": # no data? skip it. |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
45 continue |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
46 name_list.append(name) |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
47 name = dir + os.sep + name |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
48 writeFile(name, data) |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
49 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
50 # Write new file names to a file. |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
51 name_list_location = dir + os.sep + "hmmlist.txt" |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
52 with open(name_list_location, 'w') as f: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
53 for name in name_list: |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
54 f.write(name + "\n") |