0
|
1 #!/usr/bin/env python
|
|
2 # encoding: utf-8
|
|
3 '''
|
|
4 Module wrapping the NIST MSSEARCH application for matching
|
|
5 spectra one or more spectra libraries.
|
|
6 '''
|
|
7 import csv
|
|
8 import sys
|
|
9 import fileinput
|
|
10 import urllib2
|
|
11 import time
|
|
12 import utils
|
|
13 import uuid
|
|
14 import os
|
|
15 import subprocess
|
|
16 from report_generator import ReportGenerator
|
|
17
|
|
18 __author__ = "Pieter Lukasse"
|
|
19 __contact__ = "pieterlukasse@gmail.com,pieter.lukasse@wur.nl"
|
|
20 __copyright__ = "Copyright, 2015"
|
|
21 __license__ = "Apache v2"
|
|
22
|
|
23
|
|
24
|
7
|
25 def _prepare_NIST(uuid_value, nist_home_dir, nist_ini_file, spectrum_file, is_wine, log_file):
|
0
|
26 '''
|
|
27 executes the following steps:
|
|
28 - copy nist_home_dir folder to nist_home_dir+_uuid
|
|
29 - copy spectrum_file.msp to ~/.wine/drive_c/NIST_uid
|
|
30 - creates nist_home_dir+_uuid/MSSEARCH/AUTOIMP.MSD -> pointing to C:\NIST_uid\MSSEARCH\temp.msd (in case of is_wine) or to nist_home_dir+_uuidM\SSEARCH\temp.msd
|
|
31 - creates nist_home_dir+_uuid/MSSEARCH/temp.msd -> pointing to C:\NIST_uid\spectrum_file.msp (in case of is_wine) or to nist_home_dir+_uuid\spectrum_file.msp
|
|
32 and the text "10 724" in the second row
|
|
33 - copy nistms.INI to nist_home_dir+_uuid/MSSEARCH , overwriting the existing one
|
|
34 - in case of is_wine:
|
|
35 replace all occurrences of C:\NIST with C:\NIST_uid in this new nistms.INI
|
|
36 else:
|
|
37 replace all occurrences of C:\NIST with nist_home_dir+_uuid in this new nistms.INI
|
|
38 '''
|
|
39 if nist_home_dir.endswith("/") or nist_home_dir.endswith("\\"):
|
|
40 nist_home_dir = nist_home_dir[:-1]
|
|
41
|
|
42 # small validation for wine scenario
|
|
43 if is_wine and not nist_home_dir.endswith("drive_c/NIST"):
|
|
44 raise Exception('Error: invalid NIST home. For wine usage NIST home dir must be in the .wine folder and then in drive_c/NIST')
|
|
45
|
|
46 new_nist_home = nist_home_dir+uuid_value
|
|
47 utils.copy_dir(nist_home_dir, new_nist_home)
|
|
48
|
|
49 utils.copy_file(spectrum_file, new_nist_home+"/spectrum_file.msp")
|
|
50
|
9
|
51 utils.log_message(log_file, "configuring NIST input...")
|
0
|
52 # remove old file:
|
|
53 os.remove(new_nist_home+"/MSSEARCH/AUTOIMP.MSD")
|
|
54 with open(new_nist_home + "/MSSEARCH/AUTOIMP.MSD", "a") as text_file:
|
|
55 if is_wine:
|
|
56 text_file.write("C:\\NIST" + uuid_value + "\\MSSEARCH\\temp.msd")
|
|
57 else:
|
|
58 text_file.write(new_nist_home + "\\MSSEARCH\\temp.msd")
|
|
59
|
|
60 with open(new_nist_home + "/MSSEARCH/temp.msd", "a") as text_file:
|
|
61 if is_wine:
|
|
62 text_file.write("C:\\NIST" + uuid_value + "\\spectrum_file.msp\n")
|
|
63 else:
|
|
64 text_file.write(new_nist_home + "\\spectrum_file.msp\n")
|
|
65 text_file.write("10 724")
|
|
66
|
|
67 replacement_text = new_nist_home
|
|
68 if is_wine:
|
|
69 replacement_text = "C:\\NIST" + uuid_value
|
|
70
|
|
71 # remove old file
|
|
72 os.remove(new_nist_home+"/MSSEARCH/nistms.INI")
|
|
73 # make new one
|
9
|
74 utils.log_message(log_file, "configuring NIST ini...")
|
0
|
75 o = open(new_nist_home+"/MSSEARCH/nistms.INI","a") #open for append
|
|
76 # TODO : this loop/replace below is a bit limited to specific variables...either test different NIST versions or make more generic (harder in case of wine, or we need extra "home in .INI file" parameter):
|
|
77 for line in open(nist_ini_file):
|
|
78 if "Library Directory=" in line:
|
|
79 line = "Library Directory="+ new_nist_home + "\\MSSEARCH\\\n"
|
|
80 if "Dir=" in line:
|
|
81 line = "Dir="+ replacement_text + "\\MSSEARCH\\\n"
|
|
82
|
|
83 o.write(line)
|
|
84 o.close()
|
|
85
|
|
86 return new_nist_home
|
|
87
|
16
|
88 def _clean_up_NIST(new_nist_home):
|
|
89 '''
|
|
90 remove folder
|
|
91 '''
|
|
92 utils.remove_dir(new_nist_home)
|
|
93
|
14
|
94 def _run_NIST(new_nist_home, output_file, is_wine, log_file, job_size):
|
0
|
95 '''
|
|
96 - run : (wine) new_nist_home/MSSEARCH/nistms$.exe /INSTRUMENT /PAR=2
|
|
97 - monitor : new_nist_home/MSSEARCH/SRCREADY.TXT for content = "1"
|
|
98 - when ready:
|
|
99 > copy SRCRESLT.TXT to output_file
|
|
100 > kill nist process
|
|
101 > (optional)remove ~/.wine/drive_c/NIST_uid/
|
|
102 > finish
|
|
103 '''
|
|
104 # to avoid conflicts in the orphan process killing (see end of this method), we will
|
|
105 # only run NIST again after previous nistms.exe process has been killed:
|
|
106 # TODO : solution is currently only for wine (in the windows context the solution is not there yet, but parallel calls are not expected as in windows we only run tests one by one for now)
|
|
107
|
|
108
|
|
109 # remove old file, if it is there:
|
|
110 file_to_monitor = new_nist_home+"/MSSEARCH/SRCREADY.TXT"
|
|
111 if os.path.exists(file_to_monitor):
|
|
112 os.remove(file_to_monitor)
|
|
113
|
|
114 exec_path = new_nist_home + "/MSSEARCH/nistms$.exe"
|
|
115
|
|
116 pro = ""
|
|
117 if is_wine:
|
9
|
118 utils.log_message(log_file, "calling wine with " + exec_path)
|
0
|
119 cmd = ["wine "+ exec_path + " /INSTRUMENT /PAR=2"]
|
|
120 # The os.setsid() is passed in the argument preexec_fn so
|
|
121 # it's run after the fork() and before exec() to run the shell.
|
|
122 pro = subprocess.Popen(cmd, stdout=subprocess.PIPE,
|
|
123 shell=True, preexec_fn=os.setsid)
|
|
124 else:
|
|
125 cmd = [
|
|
126 exec_path,
|
|
127 "/INSTRUMENT",
|
|
128 "/PAR=2"]
|
|
129 subprocess.call(cmd)
|
|
130
|
|
131
|
5
|
132 timeSleeping = 0
|
18
|
133 startUpTime = 60
|
0
|
134 # monitor process by checking state file:
|
9
|
135 utils.log_message(log_file, "monitoring SRCREADY.TXT...")
|
14
|
136 while True:
|
0
|
137 # check if SRCREADY.TXT is there already:
|
|
138 if os.path.exists(file_to_monitor):
|
|
139 break
|
|
140 time.sleep(2)
|
10
|
141 timeSleeping = timeSleeping+2
|
14
|
142 if timeSleeping > (startUpTime + job_size*2):
|
|
143 # abort with timeout:
|
17
|
144 utils.log_message(log_file, "No results after " + str(timeSleeping) + " seconds...")
|
14
|
145 raise Exception('Error: timeout waiting for NIST results.')
|
0
|
146
|
17
|
147 utils.log_message(log_file, "done...")
|
0
|
148 # kill process:
|
|
149 #p.terminate() - not needed, nistm$ will terminate...nistms.exe is the one that
|
|
150 #stays open...and orphan..killing it:
|
|
151
|
|
152 if is_wine:
|
|
153 # pid = utils.get_process_pid("nistms.exe")
|
|
154 # os.kill(pid, 9)
|
16
|
155 os.killpg(pro.pid, 9)
|
0
|
156 else:
|
|
157 # windows case:
|
|
158 proc_name = "nistms.exe"
|
|
159 os.system("taskkill /f /im " + proc_name)
|
|
160
|
|
161 # copy SRCRESLT.TXT to output_file
|
|
162 result_file = new_nist_home+"/MSSEARCH/SRCRESLT.TXT"
|
|
163 utils.copy_file(result_file, output_file)
|
|
164
|
|
165
|
|
166 def _create_html_report(output_html_report, output_html_report_files_path, hits_dict, spectra_dict):
|
|
167 '''
|
|
168 This report will contain a page that displays essentially the same list as found in the
|
|
169 tabular output file (rendered with datatables jquery plugin), with some extra features:
|
|
170 - when user clicks on an entry, it should display the query spectrum and the hit spectrum
|
|
171 in "head to tail" and "difference" mode (see galaxy/report_example.png)
|
|
172 -> the query spectrum can be generated from the data in the input MSP file
|
|
173 -> the library "online representative" spectrum can be generated from data returned by http://webbook.nist.gov/cgi/cbook.cgi?JCAMP=C537268&Index=0&Type=Mass ,
|
|
174 where C537268 in this case is the CAS ID without the '-' separators
|
|
175 '''
|
|
176 # step 1 : generate HTML via the jinja template engine
|
|
177 # step 1.1: make sure to link the query spectrum data to the corresponding html object for quick rendering when needed
|
14
|
178
|
|
179 # copy necessary .js files as well:
|
|
180 templates_folder = os.path.dirname(__file__) + '/templates/'
|
|
181 utils.copy_dir(templates_folder, output_html_report_files_path)
|
0
|
182
|
|
183 html_file = open(output_html_report,'w')
|
|
184 html_render = ReportGenerator(os.path.dirname(__file__), 'templates/main_template.html',hits_dict, spectra_dict)
|
|
185 html_render.render(html_file)
|
|
186
|
|
187 html_file.close()
|
|
188
|
|
189
|
|
190
|
|
191
|
|
192
|
|
193 return None
|
|
194
|
|
195 def _get_extra_info_and_link_cols(data_found, data_type_found, query_link):
|
|
196 '''
|
|
197 This method will go over the data found and will return a
|
|
198 list with the following items:
|
|
199 - Experiment details where hits have been found :
|
|
200 'organism', 'tissue','experiment_name','user_name','column_type'
|
|
201 - Link that executes same query
|
|
202
|
|
203 '''
|
|
204 # set() makes a unique list:
|
|
205 organism_set = []
|
|
206 tissue_set = []
|
|
207 experiment_name_set = []
|
|
208 user_name_set = []
|
|
209 column_type_set = []
|
|
210 cas_nr_set = []
|
|
211
|
|
212 if 'organism' in data_found:
|
|
213 organism_set = set(data_found['organism'])
|
|
214 if 'tissue' in data_found:
|
|
215 tissue_set = set(data_found['tissue'])
|
|
216 if 'experiment_name' in data_found:
|
|
217 experiment_name_set = set(data_found['experiment_name'])
|
|
218 if 'user_name' in data_found:
|
|
219 user_name_set = set(data_found['user_name'])
|
|
220 if 'column_type' in data_found:
|
|
221 column_type_set = set(data_found['column_type'])
|
|
222 if 'CAS' in data_found:
|
|
223 cas_nr_set = set(data_found['CAS'])
|
|
224
|
|
225
|
|
226 result = [data_type_found,
|
|
227
|
|
228 #To let Excel interpret as link, use e.g. =HYPERLINK("http://stackoverflow.com", "friendly name"):
|
|
229 "=HYPERLINK(\""+ query_link + "\", \"Link to entries found in DB \")"]
|
|
230 return result
|
|
231
|
|
232
|
|
233
|
|
234
|
|
235
|
|
236 # alternative: ?
|
|
237 # s = requests.Session()
|
|
238 # s.verify = False
|
|
239 # #s.auth = (token01, token02)
|
|
240 # resp = s.get(url, params={'name': 'anonymous'}, stream=True)
|
|
241 # content = resp.content
|
|
242 # # transform to dictionary:
|
|
243
|
|
244
|
|
245
|
|
246 def _save_data(data_rows, headers, out_csv):
|
|
247 '''
|
|
248 Writes tab-separated data to file
|
|
249 @param data_rows: dictionary containing merged/enriched dataset
|
|
250 @param out_csv: output csv file
|
|
251 '''
|
|
252
|
|
253 # Open output file for writing
|
|
254 outfile_single_handle = open(out_csv, 'wb')
|
|
255 output_single_handle = csv.writer(outfile_single_handle, delimiter="\t")
|
|
256
|
|
257 # Write headers
|
|
258 output_single_handle.writerow(headers)
|
|
259
|
|
260 # Write one line for each row
|
|
261 for data_row in data_rows:
|
|
262 output_single_handle.writerow(data_row)
|
|
263
|
|
264 def _get_metexp_URL(metexp_dblink_file):
|
|
265 '''
|
|
266 Read out and return the URL stored in the given file.
|
|
267 '''
|
|
268 file_input = fileinput.input(metexp_dblink_file)
|
|
269 try:
|
|
270 for line in file_input:
|
|
271 if line[0] != '#':
|
|
272 # just return the first line that is not a comment line:
|
|
273 return line
|
|
274 finally:
|
|
275 file_input.close()
|
|
276
|
|
277
|
|
278 def main():
|
|
279 '''
|
|
280 Wrapper main function
|
|
281
|
|
282 The input expected is:
|
|
283 NIST_HOME dir
|
|
284 nistms.INI
|
|
285 spectrum_file.msp
|
|
286 outputfileName
|
|
287 (optional) htmlReportFile
|
|
288 (optional) htmlReportFile.files_path
|
|
289 '''
|
|
290 seconds_start = int(round(time.time()))
|
|
291
|
|
292 nist_home_dir = sys.argv[1]
|
|
293 nist_ini_file = sys.argv[2]
|
|
294 spectrum_file = sys.argv[3]
|
|
295 nist_output_file = sys.argv[4]
|
|
296 final_output_file = sys.argv[5]
|
7
|
297 output_log_file = sys.argv[6]
|
0
|
298 # html report pars:
|
|
299 output_html_report = None
|
|
300 output_html_report_files_path = None
|
7
|
301 if len(sys.argv) > 7:
|
|
302 output_html_report = sys.argv[7]
|
|
303 output_html_report_files_path = sys.argv[8]
|
0
|
304
|
|
305 is_wine = False
|
|
306 if "wine" in nist_home_dir:
|
|
307 is_wine = True
|
|
308
|
|
309 uuid_value = str(uuid.uuid4())
|
9
|
310
|
0
|
311 # prepare NIST environment for running:
|
9
|
312 new_nist_home = _prepare_NIST(uuid_value, nist_home_dir, nist_ini_file, spectrum_file, is_wine, output_log_file)
|
0
|
313
|
|
314 # run NIST search command:
|
14
|
315 spectra_dict = utils.get_spectra_file_as_dict(spectrum_file)
|
|
316 job_size = len(spectra_dict)
|
|
317 _run_NIST(new_nist_home, nist_output_file, is_wine, output_log_file, job_size)
|
0
|
318
|
16
|
319 # clean-up NIST environment:
|
|
320 _clean_up_NIST(new_nist_home)
|
|
321
|
0
|
322 # write output tabular:
|
|
323 hits_dict = utils.get_nist_out_as_dict(nist_output_file)
|
|
324 utils.save_dict_as_tsv(hits_dict, final_output_file)
|
|
325
|
|
326 # create report:
|
14
|
327 if len(sys.argv) > 7:
|
0
|
328 _create_html_report(output_html_report, output_html_report_files_path, hits_dict, spectra_dict)
|
|
329
|
|
330
|
|
331 #_save_data(enriched_data, headers, output_result)
|
|
332
|
|
333 seconds_end = int(round(time.time()))
|
|
334 print "Took " + str(seconds_end - seconds_start) + " seconds"
|
|
335
|
|
336
|
|
337
|
|
338 if __name__ == '__main__':
|
|
339 main()
|