Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics
comparison phyloconversion/prune_phytab_using_list.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 |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:5b9a38ec4a39 |
---|---|
1 #!/usr/bin/python -tt | |
2 | |
3 ##usage: ./pullgoodseqs.py <originalphytab> <bls2remove> <keep|discard> > outfile | |
4 #import modules | |
5 import sys, os, numpy, re | |
6 | |
7 def read(filename): | |
8 f = open(filename) | |
9 bad = open(sys.argv[2]) | |
10 lines = f.readlines() | |
11 badlines = bad.readlines() | |
12 badstripped = [line[:-1] for line in badlines] | |
13 str1 = '|'.join(badstripped) | |
14 str2 = '('+str1[:-1]+')' | |
15 pattern = re.compile(str2) | |
16 count=0 | |
17 for line in lines: | |
18 # line.strip() | |
19 match = pattern.findall(line) | |
20 if match and sys.argv[3] == 'keep': | |
21 print line, | |
22 if not match and sys.argv[3] == 'discard': | |
23 print line, | |
24 bad.close() | |
25 f.close() | |
26 | |
27 def main(): | |
28 read(sys.argv[1]) | |
29 | |
30 if __name__ == '__main__': | |
31 main() |