|
0
|
1 #!/usr/bin/env python
|
|
|
2
|
|
|
3 import sys, re
|
|
|
4 import os
|
|
|
5 import tempfile
|
|
|
6 import shutil, subprocess, glob
|
|
|
7 import optparse
|
|
|
8 from os.path import basename
|
|
|
9 import zipfile, tarfile, gzip
|
|
|
10 from galaxy.datatypes.checkers import *
|
|
|
11 from stacks import *
|
|
|
12
|
|
|
13 """
|
|
|
14
|
|
|
15 Created by Yvan Le Bras
|
|
|
16 yvan.le_bras@irisa.fr
|
|
|
17
|
|
|
18 Last modifications : 02/17/2014
|
|
|
19
|
|
|
20 WARNING :
|
|
|
21
|
|
|
22 STACKS_sort_read_pairs.py needs:
|
|
|
23
|
|
|
24 - STACKS scripts in your $PATH
|
|
|
25
|
|
|
26 These scripts are available after compiling the sources of STACKS :
|
|
|
27
|
|
|
28 http://creskolab.uoregon.edu/stacks/
|
|
|
29
|
|
|
30 or with the galaxy_stacks package in the Genouest toolshed
|
|
|
31
|
|
|
32
|
|
|
33 """
|
|
|
34
|
|
|
35 def __main__():
|
|
|
36
|
|
|
37
|
|
|
38 # create the working dir
|
|
|
39 os.mkdir("sort_read_outputs")
|
|
|
40 os.mkdir("assembly_outputs")
|
|
|
41 os.mkdir("samples_inputs")
|
|
|
42 os.mkdir("stacks_inputs")
|
|
|
43
|
|
|
44 # arguments recuperation
|
|
|
45 parser = optparse.OptionParser()
|
|
|
46 parser.add_option("-a")
|
|
|
47 parser.add_option("-e")
|
|
|
48 parser.add_option("-b")
|
|
|
49 parser.add_option("-c")
|
|
|
50 parser.add_option("-d")
|
|
|
51 parser.add_option("-o")
|
|
|
52 (options, args) = parser.parse_args()
|
|
|
53
|
|
|
54 # edit the command line
|
|
|
55 cmd_line1 = ["sort_read_pairs.pl"]
|
|
|
56
|
|
|
57 #parse config files and create symlink if individual files are selected
|
|
|
58
|
|
|
59 # STACKS_archive
|
|
|
60 # check if zipped files are into the tab
|
|
|
61 extract_compress_files(options.a, os.getcwd()+"/stacks_inputs")
|
|
|
62
|
|
|
63 # samples_archive
|
|
|
64 # check if zipped files are into the tab and change tab content
|
|
|
65 extract_compress_files(options.e, os.getcwd()+"/samples_inputs")
|
|
|
66
|
|
|
67 # create the sort_read_pairs command input line
|
|
|
68 cmd_line1.extend(["-p", "stacks_inputs", "-s", "samples_inputs", "-o", "sort_read_outputs"])
|
|
|
69
|
|
|
70 if options.b:
|
|
|
71 cmd_line1.extend(["-w", options.b])
|
|
|
72 if options.c:
|
|
|
73 cmd_line1.extend(["-r", options.c])
|
|
|
74
|
|
|
75 # exec command line 1
|
|
|
76 p1 = subprocess.Popen(cmd_line1)
|
|
|
77 p1.communicate()
|
|
|
78
|
|
|
79 # parse all files list and remove empty files from the output dir
|
|
|
80 for fasta_file in glob.glob("sort_read_outputs/*"):
|
|
|
81 if os.stat(fasta_file).st_size==0:
|
|
|
82 print "File "+fasta_file+" is empty"
|
|
|
83 os.remove(fasta_file)
|
|
|
84
|
|
|
85
|
|
|
86 # create the exec_velvet command input line
|
|
|
87 cmd_line2 = ["exec_velvet.pl"]
|
|
|
88 cmd_line2.extend(["-s", "sort_read_outputs", "-o", "assembly_outputs"])
|
|
|
89 cmd_line2.append("-c")
|
|
|
90
|
|
|
91 if options.d:
|
|
|
92 cmd_line2.extend(["-M", options.d])
|
|
|
93
|
|
|
94 # version
|
|
|
95 #cmd = 'sort_read_pairs.pl'+cmd_files+" "+cmd_options+" 2>&1"
|
|
|
96 #cmd2 = 'exec_velvet.pl'+cmd_files2+" -c -e /softs/local/velvet/velvet_1.2.03/ "+cmd_options2+" 2>&1"
|
|
|
97
|
|
|
98 # launch the command line 2
|
|
|
99 p2 = subprocess.Popen(cmd_line2)
|
|
|
100 p2.communicate()
|
|
|
101
|
|
|
102 # get collated.fa file
|
|
|
103 try:
|
|
|
104 shutil.copy("assembly_outputs/collated.fa", options.o)
|
|
|
105 except:
|
|
|
106 print "No result file"
|
|
|
107 sys.exit(1)
|
|
|
108
|
|
|
109 if __name__ == "__main__": __main__()
|
