comparison SMART/DiffExpAnal/countNumber_parallel.py @ 18:94ab73e8a190

Uploaded
author m-zytnicki
date Mon, 29 Apr 2013 03:20:15 -0400
parents
children
comparison
equal deleted inserted replaced
17:b0e8584489e6 18:94ab73e8a190
1 #! /usr/bin/env python
2
3
4 import optparse, os, sys, subprocess, tempfile, shutil, tarfile, random
5 from optparse import OptionParser
6
7 def stop_err(msg):
8 sys.stderr.write('%s\n' % msg)
9 sys.exit()
10
11 def toTar(tarFileName, outCountNames):
12 dir = os.path.dirname(tarFileName)
13 tfile = tarfile.open(tarFileName + ".tmp.tar", "w")
14 currentPath = os.getcwd()
15 os.chdir(dir)
16 for file in outCountNames:
17 relativeFileName = os.path.basename(file)
18 tfile.add(relativeFileName)
19 os.system("mv %s %s" % (tarFileName + ".tmp.tar", tarFileName))
20 tfile.close()
21 os.chdir(currentPath)
22
23
24 def __main__():
25 #Parse Command Line
26 parser = optparse.OptionParser()
27 parser.add_option("-i", "--input", dest="inputFile", help="input txt file, a list of overlapping results files.")
28 parser.add_option("-o", "--output", dest="outputFile", help="Out txt file.")
29 parser.add_option("-t", "--tar", dest="outputTar", default=None, help="output all count results in a tar file.")
30 (options, args) = parser.parse_args()
31
32 #Parse the input txt file and read a list of transcripts files.
33 file = open(options.inputFile, "r")
34 lines = file.readlines()
35 inputFileNames = []
36 outCountNames = []
37 outputName = options.outputFile
38 resDirName = os.path.dirname(outputName) + '/'
39
40 #Write output txt file and define all output count file names
41 out = open(outputName, "w")
42 out.write("label\tfiles\tgroup\n")
43 for line in lines:
44 tab = line.split()
45 inputFileNames.append(tab[1])
46 outCountName = resDirName + tab[0] + "_outCount_%s.csv" % random.randrange(0, 10000)
47 outCountNames.append(outCountName)
48 out.write(tab[0] + '\t' + outCountName + '\t' + tab[0][5] + '\n')
49 file.close()
50 out.close()
51
52 #Construct the lines commands
53 cmds = []
54 for i in range(len(inputFileNames)):
55 cmd = "perl %s/SMART/DiffExpAnal/countNumber.pl " % os.environ["REPET_PATH"]
56 opts = "%s %s " % (inputFileNames[i], outCountNames[i])
57 cmd += opts
58 cmds.append(cmd)
59
60 tmp_files = []
61 for i in range(len(cmds)):
62 try:
63 tmp_out = tempfile.NamedTemporaryFile().name
64 tmp_files.append(tmp_out)
65 tmp_stdout = open(tmp_out, 'wb')
66 tmp_err = tempfile.NamedTemporaryFile().name
67 tmp_files.append(tmp_err)
68 tmp_stderr = open(tmp_err, 'wb')
69 proc = subprocess.Popen(args=cmds[i], shell=True, cwd=".", stdout=tmp_stdout, stderr=tmp_stderr)
70 returncode = proc.wait()
71 tmp_stderr.close()
72 #get stderr, allowing for case where it's very large
73 tmp_stderr = open(tmp_err, 'rb')
74 stderr = ''
75 buffsize = 1048576
76 try:
77 while True:
78 stderr += tmp_stderr.read(buffsize)
79 if not stderr or len(stderr) % buffsize != 0:
80 break
81 except OverflowError:
82 pass
83 tmp_stdout.close()
84 tmp_stderr.close()
85 if returncode != 0:
86 raise Exception, stderr
87 except Exception, e:
88 stop_err('Error in :\n' + str(e))
89
90 if options.outputTar != None:
91 toTar(options.outputTar, outCountNames)
92
93 for tmp_file in tmp_files:
94 os.remove(tmp_file)
95
96 if __name__=="__main__":__main__()