Mercurial > repos > nml > stringmlst

comparison split_by_allele.py @ 1:4e03573653fe draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 008f4667b70be22e9ddf496738b3f74bb942ed28
author nml
date Tue, 19 Sep 2017 16:34:57 -0400
parents fc0f15ca12e0
children
comparison
equal deleted inserted replaced
0:fc0f15ca12e0 1:4e03573653fe
1 #!/usr/bin/env python 1 #!/usr/bin/env python
2 import getopt 2 import getopt
3 import os
3 import sys 4 import sys
4 import os 5
5 from Bio import SeqIO 6 from Bio import SeqIO
6 7
7 def split_allele_file(alleles,profiles): 8 ERROR_MSG = "Error could not parse out allele name and number from '%s'"
8 9
10
11 def split_allele_file(alleles, profiles):
12
9 writers = {} 13 writers = {}
10 14
11 handle = open(alleles, "rU") 15 handle = open(alleles, "rU")
12 for record in SeqIO.parse(handle, "fasta"): 16 for record in SeqIO.parse(handle, "fasta"):
13 17
14 seqid=record.id 18 seqid = record.id
15 19
16 #split out the alelle name from the version number 20 # split out the alelle name from the version number
17 #attempting to split based on '-' first, if that fails, then '_' 21 # attempting to split based on '-' first, if that fails, then '_'
18 result = seqid.split('_') 22 result = seqid.split('_')
19 23
20 if len(result) !=2: 24 if len(result) != 2:
21 result = seqid.split('-') 25 result = seqid.split('-')
22 if len(result) ==2: 26 if len(result) == 2:
23 newid = '_'.join(result) 27 newid = '_'.join(result)
24 record.id = newid 28 record.id = newid
25 else: 29 else:
26 print "Error could not parse out allele name and number from '%s'" % seqid 30 print(ERROR_MSG % seqid)
27 exit(0) 31 exit(0)
28
29
30 name,num = result
31 32
33 name, num = result
32 34
33 #if writer exist, then write to that current fasta file 35 # if writer exist, then write to that current fasta file
34 if name in writers: 36 if name in writers:
35 SeqIO.write(record, writers[name], "fasta") 37 SeqIO.write(record, writers[name], "fasta")
36 else: 38 else:
37 #new allele found, create new writer and add the first record 39 # new allele found, create new writer and add the first record
38 file_name = name + '.fasta' 40 file_name = name + '.fasta'
39 output_fh = open(file_name, "w") 41 output_fh = open(file_name, "w")
40 SeqIO.write(record, output_fh, "fasta") 42 SeqIO.write(record, output_fh, "fasta")
41 writers[name] = output_fh 43 writers[name] = output_fh
42 44
43 handle.close() 45 handle.close()
44 46
45 #creat config file based on the alleles found 47 # create config file based on the alleles found
46 with open('config.txt','w') as cfile: 48 with open('config.txt', 'w') as cfile:
47 cfile.write("[loci]\n") 49 cfile.write("[loci]\n")
48 for name, writer in writers.iteritems() : 50 for name, writer in writers.items():
49 path = os.path.realpath(writer.name) 51 path = os.path.realpath(writer.name)
50 cfile.write("%s\t%s\n" % (name,path)) 52 cfile.write("%s\t%s\n" % (name, path))
51 cfile.write("[profile]\n") 53 cfile.write("[profile]\n")
52 cfile.write("profile\t%s\n" % profiles) 54 cfile.write("profile\t%s\n" % profiles)
53 55
54
55 return 56 return
56 57
57 58
58 alleles=None 59 alleles = None
59 profiles=None 60 profiles = None
60 61
61 """Input arguments""" 62 """Input arguments"""
62 options, remainder = getopt.getopt(sys.argv[1:], '', [ 63 options, remainder = getopt.getopt(sys.argv[1:], '', [
63 'alleles=', 64 'alleles=',
64 'profiles=' 65 'profiles='
69 alleles = arg 70 alleles = arg
70 elif opt in ('--profiles'): 71 elif opt in ('--profiles'):
71 profiles = arg 72 profiles = arg
72 73
73 if alleles and profiles: 74 if alleles and profiles:
74 split_allele_file(alleles,profiles) 75 split_allele_file(alleles, profiles)
75