|
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
|
|
14
|
88 def _run_NIST(new_nist_home, output_file, is_wine, log_file, job_size):
|
|
0
|
89 '''
|
|
|
90 - run : (wine) new_nist_home/MSSEARCH/nistms$.exe /INSTRUMENT /PAR=2
|
|
|
91 - monitor : new_nist_home/MSSEARCH/SRCREADY.TXT for content = "1"
|
|
|
92 - when ready:
|
|
|
93 > copy SRCRESLT.TXT to output_file
|
|
|
94 > kill nist process
|
|
|
95 > (optional)remove ~/.wine/drive_c/NIST_uid/
|
|
|
96 > finish
|
|
|
97 '''
|
|
|
98 # to avoid conflicts in the orphan process killing (see end of this method), we will
|
|
|
99 # only run NIST again after previous nistms.exe process has been killed:
|
|
|
100 # TODO : solution is currently only for wine (in the windows context the solution is not there yet, but parallel calls are not expected as in windows we only run tests one by one for now)
|
|
|
101 # if is_wine:
|
|
|
102 # while True:
|
|
|
103 # # check if process exists. If not, break loop and continue
|
|
|
104 # pid = utils.get_process_pid("nistms.exe")
|
|
|
105 # if pid == -1:
|
|
|
106 # break
|
|
|
107 # time.sleep(2)
|
|
|
108
|
|
|
109
|
|
|
110 # remove old file, if it is there:
|
|
|
111 file_to_monitor = new_nist_home+"/MSSEARCH/SRCREADY.TXT"
|
|
|
112 if os.path.exists(file_to_monitor):
|
|
|
113 os.remove(file_to_monitor)
|
|
|
114
|
|
|
115 exec_path = new_nist_home + "/MSSEARCH/nistms$.exe"
|
|
|
116
|
|
|
117 pro = ""
|
|
|
118 if is_wine:
|
|
9
|
119 utils.log_message(log_file, "calling wine with " + exec_path)
|
|
0
|
120 cmd = ["wine "+ exec_path + " /INSTRUMENT /PAR=2"]
|
|
|
121 # The os.setsid() is passed in the argument preexec_fn so
|
|
|
122 # it's run after the fork() and before exec() to run the shell.
|
|
|
123 pro = subprocess.Popen(cmd, stdout=subprocess.PIPE,
|
|
|
124 shell=True, preexec_fn=os.setsid)
|
|
|
125 else:
|
|
|
126 cmd = [
|
|
|
127 exec_path,
|
|
|
128 "/INSTRUMENT",
|
|
|
129 "/PAR=2"]
|
|
|
130 subprocess.call(cmd)
|
|
|
131
|
|
|
132
|
|
5
|
133 timeSleeping = 0
|
|
14
|
134 startUpTime = 30
|
|
0
|
135 # monitor process by checking state file:
|
|
9
|
136 utils.log_message(log_file, "monitoring SRCREADY.TXT...")
|
|
14
|
137 # TODO2: do the controller part
|
|
|
138 # TODO3: do the popover part
|
|
|
139 while True:
|
|
0
|
140 # check if SRCREADY.TXT is there already:
|
|
|
141 if os.path.exists(file_to_monitor):
|
|
|
142 break
|
|
|
143 time.sleep(2)
|
|
10
|
144 timeSleeping = timeSleeping+2
|
|
14
|
145 if timeSleeping > (startUpTime + job_size*2):
|
|
|
146 # abort with timeout:
|
|
|
147 utils.log_message(log_file, "No results after " + timeSleeping + " seconds...")
|
|
|
148 raise Exception('Error: timeout waiting for NIST results.')
|
|
0
|
149
|
|
|
150 # kill process:
|
|
|
151 #p.terminate() - not needed, nistm$ will terminate...nistms.exe is the one that
|
|
|
152 #stays open...and orphan..killing it:
|
|
|
153
|
|
|
154 if is_wine:
|
|
|
155 # pid = utils.get_process_pid("nistms.exe")
|
|
|
156 # os.kill(pid, 9)
|
|
|
157 os.killpg(pro.pid, 9)
|
|
|
158 else:
|
|
|
159 # windows case:
|
|
|
160 proc_name = "nistms.exe"
|
|
|
161 os.system("taskkill /f /im " + proc_name)
|
|
|
162
|
|
|
163 # copy SRCRESLT.TXT to output_file
|
|
|
164 result_file = new_nist_home+"/MSSEARCH/SRCRESLT.TXT"
|
|
|
165 utils.copy_file(result_file, output_file)
|
|
|
166
|
|
|
167
|
|
|
168 def _create_html_report(output_html_report, output_html_report_files_path, hits_dict, spectra_dict):
|
|
|
169 '''
|
|
|
170 This report will contain a page that displays essentially the same list as found in the
|
|
|
171 tabular output file (rendered with datatables jquery plugin), with some extra features:
|
|
|
172 - when user clicks on an entry, it should display the query spectrum and the hit spectrum
|
|
|
173 in "head to tail" and "difference" mode (see galaxy/report_example.png)
|
|
|
174 -> the query spectrum can be generated from the data in the input MSP file
|
|
|
175 -> 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 ,
|
|
|
176 where C537268 in this case is the CAS ID without the '-' separators
|
|
|
177 '''
|
|
|
178 # step 1 : generate HTML via the jinja template engine
|
|
|
179 # step 1.1: make sure to link the query spectrum data to the corresponding html object for quick rendering when needed
|
|
14
|
180
|
|
|
181 # copy necessary .js files as well:
|
|
|
182 templates_folder = os.path.dirname(__file__) + '/templates/'
|
|
|
183 utils.copy_dir(templates_folder, output_html_report_files_path)
|
|
0
|
184
|
|
|
185 html_file = open(output_html_report,'w')
|
|
|
186 html_render = ReportGenerator(os.path.dirname(__file__), 'templates/main_template.html',hits_dict, spectra_dict)
|
|
|
187 html_render.render(html_file)
|
|
|
188
|
|
|
189 html_file.close()
|
|
|
190
|
|
|
191
|
|
|
192
|
|
|
193
|
|
|
194
|
|
|
195 return None
|
|
|
196
|
|
|
197 def _get_extra_info_and_link_cols(data_found, data_type_found, query_link):
|
|
|
198 '''
|
|
|
199 This method will go over the data found and will return a
|
|
|
200 list with the following items:
|
|
|
201 - Experiment details where hits have been found :
|
|
|
202 'organism', 'tissue','experiment_name','user_name','column_type'
|
|
|
203 - Link that executes same query
|
|
|
204
|
|
|
205 '''
|
|
|
206 # set() makes a unique list:
|
|
|
207 organism_set = []
|
|
|
208 tissue_set = []
|
|
|
209 experiment_name_set = []
|
|
|
210 user_name_set = []
|
|
|
211 column_type_set = []
|
|
|
212 cas_nr_set = []
|
|
|
213
|
|
|
214 if 'organism' in data_found:
|
|
|
215 organism_set = set(data_found['organism'])
|
|
|
216 if 'tissue' in data_found:
|
|
|
217 tissue_set = set(data_found['tissue'])
|
|
|
218 if 'experiment_name' in data_found:
|
|
|
219 experiment_name_set = set(data_found['experiment_name'])
|
|
|
220 if 'user_name' in data_found:
|
|
|
221 user_name_set = set(data_found['user_name'])
|
|
|
222 if 'column_type' in data_found:
|
|
|
223 column_type_set = set(data_found['column_type'])
|
|
|
224 if 'CAS' in data_found:
|
|
|
225 cas_nr_set = set(data_found['CAS'])
|
|
|
226
|
|
|
227
|
|
|
228 result = [data_type_found,
|
|
|
229
|
|
|
230 #To let Excel interpret as link, use e.g. =HYPERLINK("http://stackoverflow.com", "friendly name"):
|
|
|
231 "=HYPERLINK(\""+ query_link + "\", \"Link to entries found in DB \")"]
|
|
|
232 return result
|
|
|
233
|
|
|
234
|
|
|
235
|
|
|
236
|
|
|
237
|
|
|
238 # alternative: ?
|
|
|
239 # s = requests.Session()
|
|
|
240 # s.verify = False
|
|
|
241 # #s.auth = (token01, token02)
|
|
|
242 # resp = s.get(url, params={'name': 'anonymous'}, stream=True)
|
|
|
243 # content = resp.content
|
|
|
244 # # transform to dictionary:
|
|
|
245
|
|
|
246
|
|
|
247
|
|
|
248 def _save_data(data_rows, headers, out_csv):
|
|
|
249 '''
|
|
|
250 Writes tab-separated data to file
|
|
|
251 @param data_rows: dictionary containing merged/enriched dataset
|
|
|
252 @param out_csv: output csv file
|
|
|
253 '''
|
|
|
254
|
|
|
255 # Open output file for writing
|
|
|
256 outfile_single_handle = open(out_csv, 'wb')
|
|
|
257 output_single_handle = csv.writer(outfile_single_handle, delimiter="\t")
|
|
|
258
|
|
|
259 # Write headers
|
|
|
260 output_single_handle.writerow(headers)
|
|
|
261
|
|
|
262 # Write one line for each row
|
|
|
263 for data_row in data_rows:
|
|
|
264 output_single_handle.writerow(data_row)
|
|
|
265
|
|
|
266 def _get_metexp_URL(metexp_dblink_file):
|
|
|
267 '''
|
|
|
268 Read out and return the URL stored in the given file.
|
|
|
269 '''
|
|
|
270 file_input = fileinput.input(metexp_dblink_file)
|
|
|
271 try:
|
|
|
272 for line in file_input:
|
|
|
273 if line[0] != '#':
|
|
|
274 # just return the first line that is not a comment line:
|
|
|
275 return line
|
|
|
276 finally:
|
|
|
277 file_input.close()
|
|
|
278
|
|
|
279
|
|
|
280 def main():
|
|
|
281 '''
|
|
|
282 Wrapper main function
|
|
|
283
|
|
|
284 The input expected is:
|
|
|
285 NIST_HOME dir
|
|
|
286 nistms.INI
|
|
|
287 spectrum_file.msp
|
|
|
288 outputfileName
|
|
|
289 (optional) htmlReportFile
|
|
|
290 (optional) htmlReportFile.files_path
|
|
|
291 '''
|
|
|
292 seconds_start = int(round(time.time()))
|
|
|
293
|
|
|
294 nist_home_dir = sys.argv[1]
|
|
|
295 nist_ini_file = sys.argv[2]
|
|
|
296 spectrum_file = sys.argv[3]
|
|
|
297 nist_output_file = sys.argv[4]
|
|
|
298 final_output_file = sys.argv[5]
|
|
7
|
299 output_log_file = sys.argv[6]
|
|
0
|
300 # html report pars:
|
|
|
301 output_html_report = None
|
|
|
302 output_html_report_files_path = None
|
|
7
|
303 if len(sys.argv) > 7:
|
|
|
304 output_html_report = sys.argv[7]
|
|
|
305 output_html_report_files_path = sys.argv[8]
|
|
0
|
306
|
|
|
307 is_wine = False
|
|
|
308 if "wine" in nist_home_dir:
|
|
|
309 is_wine = True
|
|
|
310
|
|
|
311 uuid_value = str(uuid.uuid4())
|
|
9
|
312
|
|
0
|
313 # prepare NIST environment for running:
|
|
9
|
314 new_nist_home = _prepare_NIST(uuid_value, nist_home_dir, nist_ini_file, spectrum_file, is_wine, output_log_file)
|
|
0
|
315
|
|
|
316 # run NIST search command:
|
|
14
|
317 spectra_dict = utils.get_spectra_file_as_dict(spectrum_file)
|
|
|
318 job_size = len(spectra_dict)
|
|
|
319 _run_NIST(new_nist_home, nist_output_file, is_wine, output_log_file, job_size)
|
|
0
|
320
|
|
|
321 # write output tabular:
|
|
|
322 hits_dict = utils.get_nist_out_as_dict(nist_output_file)
|
|
|
323 utils.save_dict_as_tsv(hits_dict, final_output_file)
|
|
|
324
|
|
|
325 # create report:
|
|
14
|
326 if len(sys.argv) > 7:
|
|
0
|
327 _create_html_report(output_html_report, output_html_report_files_path, hits_dict, spectra_dict)
|
|
|
328
|
|
|
329
|
|
|
330 #_save_data(enriched_data, headers, output_result)
|
|
|
331
|
|
|
332 seconds_end = int(round(time.time()))
|
|
|
333 print "Took " + str(seconds_end - seconds_start) + " seconds"
|
|
|
334
|
|
|
335
|
|
|
336
|
|
|
337 if __name__ == '__main__':
|
|
|
338 main()
|
