|
0
|
1 '''
|
|
|
2 Created on Mar 27, 2012
|
|
|
3
|
|
|
4 @author: marcelk
|
|
|
5 '''
|
|
|
6 from GCMS import combine_output
|
|
|
7 from pkg_resources import resource_filename # @UnresolvedImport # pylint: disable=E0611
|
|
|
8 import os
|
|
|
9 import shutil
|
|
|
10 import tempfile
|
|
|
11 import unittest
|
|
|
12
|
|
|
13
|
|
|
14 class Test(unittest.TestCase):
|
|
|
15 '''
|
|
|
16 Tests for the 'combine_output' Galaxy tool
|
|
|
17 '''
|
|
|
18
|
|
|
19 def setUp(self):
|
|
|
20 self.rf_output = resource_filename(__name__, "data/RankFilter.txt")
|
|
|
21 self.cl_output = resource_filename(__name__, "data/CasLookup.txt")
|
|
|
22
|
|
|
23 def test_process_data(self):
|
|
|
24 '''
|
|
|
25 Tests the processing of the RankFilter and CasLookup files into dictionaries
|
|
|
26 '''
|
|
|
27 rfdata = combine_output._process_data(self.rf_output)
|
|
|
28 cldata = combine_output._process_data(self.cl_output)
|
|
|
29 self.assertEqual(set([' 18457-04-0', ' 55133-95-4', ' 58-08-2', ' 112-34-5']), set(rfdata['CAS']))
|
|
|
30 self.assertEqual(set(['C58082', 'C18457040', 'C55133954', 'C112345']), set(cldata['CAS']))
|
|
|
31
|
|
|
32 def test_add_hit(self):
|
|
|
33 '''
|
|
|
34 Tests the combination of two records from both the RankFilter- and CasLookup-tools
|
|
|
35 '''
|
|
|
36 rfdata = combine_output._process_data(self.rf_output)
|
|
|
37 cldata = combine_output._process_data(self.cl_output)
|
|
|
38 index = 0
|
|
|
39 rf_record = dict(zip(rfdata.keys(), [rfdata[key][index] for key in rfdata.keys()]))
|
|
|
40 cl_record = dict(zip(cldata.keys(), [cldata[key][index] for key in cldata.keys()]))
|
|
|
41
|
|
|
42 hit = combine_output._add_hit(rf_record, cl_record)
|
|
|
43 self.assertEqual(len(hit), 27)
|
|
|
44
|
|
|
45 # Pass empty record, should fail combination
|
|
|
46 self.assertRaises(KeyError, combine_output._add_hit, rf_record, {})
|
|
|
47
|
|
|
48 def test_merge_data(self):
|
|
|
49 '''
|
|
|
50 Tests the merging of the RankFilter and CasLookup data
|
|
|
51 '''
|
|
|
52 rfdata = combine_output._process_data(self.rf_output)
|
|
|
53 cldata = combine_output._process_data(self.cl_output)
|
|
|
54 merged, _ = combine_output._merge_data(rfdata, cldata)
|
|
|
55 centrotypes = _get_centrotypes(merged)
|
|
|
56 self.failUnless(all(centrotype in centrotypes for centrotype in ('2716','12723', '3403', '12710')))
|
|
|
57
|
|
|
58 def _get_centrotypes(merged):
|
|
|
59 '''
|
|
|
60 returns centrotype codes found in merged set
|
|
|
61 '''
|
|
|
62 result = []
|
|
|
63 for item_idx in xrange(len(merged)):
|
|
|
64 item = merged[item_idx]
|
|
|
65 centrotype = item[0][0]
|
|
|
66 result.append(centrotype)
|
|
|
67
|
|
|
68 return result
|
|
|
69
|
|
|
70 def test_remove_formula(self):
|
|
|
71 '''
|
|
|
72 Tests the removal of the Formula from the 'Name' field (RankFilter output)
|
|
|
73 '''
|
|
|
74 name = "Caffeine C8H10N4O2"
|
|
|
75 compound_name, compound_formula = combine_output._remove_formula(name)
|
|
|
76 self.assertEqual(compound_name, 'Caffeine')
|
|
|
77 self.assertEqual(compound_formula, 'C8H10N4O2')
|
|
|
78 name = "Ethanol C2H6O"
|
|
|
79 compound_name, compound_formula = combine_output._remove_formula(name)
|
|
|
80 self.assertEqual(compound_name, 'Ethanol')
|
|
|
81 self.assertEqual(compound_formula, 'C2H6O')
|
|
|
82 # No formula to remove
|
|
|
83 name = "Butanoic acid, 4-[(trimethylsilyl)oxy]-, trimethylsilyl ester"
|
|
|
84 compound_name, compound_formula = combine_output._remove_formula(name)
|
|
|
85 self.assertEqual(compound_name, name)
|
|
|
86 self.assertEqual(compound_formula, False)
|
|
|
87
|
|
|
88 def test_save_data(self):
|
|
|
89 '''
|
|
|
90 Tests the creation of the output tabular files (no content testing)
|
|
|
91 '''
|
|
|
92 temp_folder = tempfile.mkdtemp(prefix='gcms_combine_output_')
|
|
|
93 saved_single_data = '{0}/{1}'.format(temp_folder, 'output_single.tsv')
|
|
|
94 saved_multi_data = '{0}/{1}'.format(temp_folder, 'output_multi.tsv')
|
|
|
95 rfdata = combine_output._process_data(self.rf_output)
|
|
|
96 cldata = combine_output._process_data(self.cl_output)
|
|
|
97 merged, nhits = combine_output._merge_data(rfdata, cldata)
|
|
|
98 combine_output._save_data(merged, nhits, saved_single_data, saved_multi_data)
|
|
|
99 self.failUnless(os.path.exists(saved_single_data))
|
|
|
100 self.failUnless(os.path.exists(saved_multi_data))
|
|
|
101 shutil.rmtree(temp_folder)
|
|
|
102
|
|
|
103
|
|
|
104 if __name__ == "__main__":
|
|
|
105 #import sys;sys.argv = ['', 'Test.testName']
|
|
|
106 unittest.main()
|
