comparison mpagenomics_normalize-7dc6ce39fb89/preprocess.py @ 0:84b13b0e2b85

Uploaded
author sblanck
date Thu, 07 May 2015 08:22:36 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:84b13b0e2b85
1 import os
2 import re
3 import shutil
4 import sys
5 import subprocess
6 import zipfile
7 import optparse
8
9 def main():
10
11 parser = optparse.OptionParser()
12 parser.add_option('-s', action="store", dest='summary')
13 parser.add_option('-e', action="store", dest='dataSetName')
14 parser.add_option('-p', action="store", dest='new_file_path')
15 parser.add_option('-c', action="store", dest='inputcdffull_name')
16 parser.add_option('-f', action="store", dest='inputufl_name')
17 parser.add_option('-g', action="store", dest='inputugp_name')
18 parser.add_option('-a', action="store", dest='inputacs_name')
19 parser.add_option('-d', action="store", dest='inputcdffull')
20 parser.add_option('-v', action="store", dest='inputufl')
21 parser.add_option('-w', action="store", dest='inputugp')
22 parser.add_option('-b', action="store", dest='inputacs')
23 parser.add_option('-t', action="store", dest='tumorcsv')
24 parser.add_option('-y', action="store", dest='settingsType')
25 parser.add_option('-o', action="store", dest='outputgraph')
26 parser.add_option('-z', action="store", dest='zipfigures')
27 parser.add_option('-k', action="store", dest='outputlog')
28 parser.add_option('-l', action="store", dest='log')
29 parser.add_option('-u', action="store", dest='user_id')
30
31 parser.add_option('-i', action="append", dest='inputFile', default=[])
32 parser.add_option('-n', action='append', dest='inputFileName', default=[])
33
34 options, args = parser.parse_args()
35
36 dataSetName=options.dataSetName
37 destinationPath=os.path.join(options.new_file_path, options.user_id, dataSetName)
38
39 mpagenomics_dir = os.path.join(options.new_file_path,"mpagenomics",options.user_id)
40 data_dir = os.path.join(options.new_file_path, options.user_id)
41
42 try:
43 os.makedirs(data_dir)
44 except:
45 shutil.rmtree(data_dir)
46 os.makedirs(data_dir)
47
48 if (not os.path.isdir(mpagenomics_dir)):
49 os.makedirs(mpagenomics_dir)
50
51 for inputFile, inputFileName in zip(options.inputFile,options.inputFileName):
52 source = inputFile
53 destination=os.path.join(data_dir,inputFileName)
54 _copy(source,destination)
55
56
57 cdffull_name=options.inputcdffull_name
58 if (cdffull_name.count(",") != 0):
59 chipType=cdffull_name.split(",",1)[0]
60 tagExt=cdffull_name.split(",",1)[1]
61 tag=tagExt.split(".",1)[0]
62 else:
63 chipType=cdffull_name.split(".",1)[0]
64 tag=""
65
66 _copy(options.inputcdffull,os.path.join(data_dir, options.inputcdffull_name))
67 _copy(options.inputugp,os.path.join(data_dir, options.inputugp_name))
68 _copy(options.inputufl,os.path.join(data_dir, options.inputufl_name))
69 _copy(options.inputacs,os.path.join(data_dir, options.inputacs_name))
70
71
72 fig_dir = os.path.join("mpagenomics", options.user_id, "figures", dataSetName, "signal")
73 abs_fig_dir = os.path.join(options.new_file_path, fig_dir)
74
75
76 retcode = _preprocess(chipType, dataSetName, mpagenomics_dir, data_dir, options.new_file_path, options.tumorcsv, options.settingsType, options.outputgraph, options.outputlog, options.log, tag)
77
78 if (retcode == 0):
79 if (os.path.isdir(abs_fig_dir)) and (options.outputgraph == "TRUE"):
80
81 new_files = os.listdir(abs_fig_dir)
82 zipbuf = zipfile.ZipFile(os.path.join(abs_fig_dir, options.zipfigures), 'w', zipfile.ZIP_DEFLATED)
83 for current_file in new_files:
84 fn = os.path.join(abs_fig_dir, current_file)
85 relfn = fn[len(abs_fig_dir) + len(os.sep):]
86 zipbuf.write(fn, relfn)
87
88 f = open(options.summary, "w")
89 # Create report
90 try:
91 for inputFileName in options.inputFileName:
92 f.write("%s\t%s\t%s\n" %(inputFileName,dataSetName,chipType))
93 finally:
94 shutil.rmtree(data_dir)
95 f.close()
96
97 sys.exit(retcode)
98
99 sys.exit(retcode)
100
101
102 def _copy(source, destination):
103 try:
104 os.symlink(source, destination)
105 except:
106 shutil.copy(source, destination)
107
108 def _preprocess (chipType,dataset,mpagenomics_dir,data_dir,tmp_dir,tumor,settingType,outputgraph,outputlog,log,tag):
109 script_dir=os.path.dirname(os.path.abspath(__file__))
110
111 if (outputlog=="TRUE"):
112 errfile=open(log,'w')
113 else:
114 errfile=open(os.path.join(tmp_dir,"errfile.log"),'w')
115
116 retcode = subprocess.call(["Rscript", os.path.join(script_dir,"preprocess.R"), chipType, dataset, mpagenomics_dir, data_dir, tumor, settingType, outputgraph, tag], stdout = errfile, stderr = errfile)
117 return(retcode)
118
119
120 if __name__ == "__main__":
121 main()