|
18
|
1 from commons.core.utils.FileUtils import FileUtils
|
|
|
2 import os
|
|
|
3 import unittest
|
|
|
4 from commons.launcher.LaunchTallymer import ConvertUtils
|
|
|
5
|
|
|
6 class Test_ConvertUtils(unittest.TestCase):
|
|
|
7
|
|
|
8 def test_convertTallymerFormatIntoMapFormatAndGenerateData(self):
|
|
|
9 inputFastaFileName = "input.fa"
|
|
|
10 inputTallymerSearchFileName = "input.tallymer"
|
|
|
11 expMapFileName = "exp.map"
|
|
|
12 obsMapFileName = "obs.map"
|
|
|
13 self._writeInputFasta(inputFastaFileName)
|
|
|
14 self._writeInputTallymerSearch(inputTallymerSearchFileName)
|
|
|
15 self._writeExpMap(expMapFileName)
|
|
|
16
|
|
|
17 expOccNb = 4
|
|
|
18 expDKmer2Occ = {"caagatgcgtaacggccata":1,
|
|
|
19 "aagatgcgtaacggccatac":1,
|
|
|
20 "agatgcgtaacggccataca":1,
|
|
|
21 "gatgcgtaacggccatacat":1}
|
|
|
22 expPlotData = {4379: 17,
|
|
|
23 4380: 17,
|
|
|
24 4381: 16,
|
|
|
25 4382: 16}
|
|
|
26
|
|
|
27 obsOccNb, obsDKmer2Occ, obsPlotData, obsPlotData2 = ConvertUtils.convertTallymerFormatIntoMapFormatAndGenerateData(inputFastaFileName, inputTallymerSearchFileName, obsMapFileName)
|
|
|
28
|
|
|
29 self.assertEquals(expOccNb, obsOccNb)
|
|
|
30 self.assertEquals(expDKmer2Occ, obsDKmer2Occ)
|
|
|
31 self.assertEquals(expPlotData, obsPlotData)
|
|
|
32 self.assertTrue(FileUtils.are2FilesIdentical(expMapFileName, obsMapFileName))
|
|
|
33
|
|
|
34 os.remove(inputFastaFileName)
|
|
|
35 os.remove(inputTallymerSearchFileName)
|
|
|
36 os.remove(expMapFileName)
|
|
|
37 os.remove(obsMapFileName)
|
|
|
38
|
|
|
39 def test_convertTallymerFormatIntoMapFormatAndGenerateDataMultiFasta(self):
|
|
|
40 inputFastaFileName = "input.fa"
|
|
|
41 inputTallymerSearchFileName = "input.tallymer"
|
|
|
42 expMapFileName = "exp.map"
|
|
|
43 obsMapFileName = "obs.map"
|
|
|
44 self._writeInputMultiFasta(inputFastaFileName)
|
|
|
45 self._writeInputTallymerSearchMultiFasta(inputTallymerSearchFileName)
|
|
|
46 self._writeExpMapMultiFasta(expMapFileName)
|
|
|
47
|
|
|
48 expOccNb = 6
|
|
|
49 expDKmer2Occ = {"caagatgcgtaacggccata":1,
|
|
|
50 "aagatgcgtaacggccatac":1,
|
|
|
51 "agatgcgtaacggccataca":1,
|
|
|
52 "gatgcgtaacggccatacat":1,
|
|
|
53 "agatgcgtaacggccataca":2,
|
|
|
54 "gatgcgtaacggccatacat":2}
|
|
|
55 expPlotData = {4379: 17, 4380: 17, 4381: 16, 4382: 35}
|
|
|
56
|
|
|
57 obsOccNb, obsDKmer2Occ, obsPlotData, obsPlotData2 = ConvertUtils.convertTallymerFormatIntoMapFormatAndGenerateData(inputFastaFileName, inputTallymerSearchFileName, obsMapFileName)
|
|
|
58
|
|
|
59 self.assertEquals(expOccNb, obsOccNb)
|
|
|
60 self.assertEquals(expDKmer2Occ, obsDKmer2Occ)
|
|
|
61 self.assertEquals(expPlotData, obsPlotData)
|
|
|
62 self.assertTrue(FileUtils.are2FilesIdentical(expMapFileName, obsMapFileName))
|
|
|
63
|
|
|
64 os.remove(inputFastaFileName)
|
|
|
65 os.remove(inputTallymerSearchFileName)
|
|
|
66 os.remove(expMapFileName)
|
|
|
67 os.remove(obsMapFileName)
|
|
|
68
|
|
|
69 def _writeInputTallymerSearch(self, tallymerFormatFileName):
|
|
|
70 with open(tallymerFormatFileName, 'w') as f:
|
|
|
71 f.write("0\t+4378\t17\tcaagatgcgtaacggccata\n")
|
|
|
72 f.write("0\t+4379\t17\taagatgcgtaacggccatac\n")
|
|
|
73 f.write("0\t+4380\t16\tagatgcgtaacggccataca\n")
|
|
|
74 f.write("0\t+4381\t16\tgatgcgtaacggccatacat\n")
|
|
|
75
|
|
|
76 def _writeInputFasta(self, fastaFileName):
|
|
|
77 with open(fastaFileName, 'w') as f:
|
|
|
78 f.write(">Seq1\n")
|
|
|
79 f.write("caagatgcgtaacggccata\n")
|
|
|
80 f.write("aagatgcgtaacggccatac\n")
|
|
|
81 f.write("agatgcgtaacggccataca\n")
|
|
|
82 f.write("gatgcgtaacggccatacat\n")
|
|
|
83
|
|
|
84 def _writeExpMap(self, mapFileName):
|
|
|
85 with open(mapFileName, 'w') as f:
|
|
|
86 f.write("caagatgcgtaacggccata_17\tSeq1\t4379\t4399\t17\n")
|
|
|
87 f.write("aagatgcgtaacggccatac_17\tSeq1\t4380\t4400\t17\n")
|
|
|
88 f.write("agatgcgtaacggccataca_16\tSeq1\t4381\t4401\t16\n")
|
|
|
89 f.write("gatgcgtaacggccatacat_16\tSeq1\t4382\t4402\t16\n")
|
|
|
90
|
|
|
91 def _writeInputTallymerSearchMultiFasta(self, tallymerFormatFileName):
|
|
|
92 with open(tallymerFormatFileName, 'w') as f:
|
|
|
93 f.write("0\t+4378\t17\tcaagatgcgtaacggccata\n")
|
|
|
94 f.write("0\t+4379\t17\taagatgcgtaacggccatac\n")
|
|
|
95 f.write("0\t+4380\t16\tagatgcgtaacggccataca\n")
|
|
|
96 f.write("0\t+4381\t16\tgatgcgtaacggccatacat\n")
|
|
|
97 f.write("1\t+4380\t16\tagatgcgtaacggccataca\n")
|
|
|
98 f.write("1\t+4381\t35\tgatgcgtaacggccatacat\n")
|
|
|
99
|
|
|
100 def _writeInputMultiFasta(self, fastaFileName):
|
|
|
101 with open(fastaFileName, 'w') as f:
|
|
|
102 f.write(">Seq1\n")
|
|
|
103 f.write("caagatgcgtaacggccata\n")
|
|
|
104 f.write("aagatgcgtaacggccatac\n")
|
|
|
105 f.write("agatgcgtaacggccataca\n")
|
|
|
106 f.write("gatgcgtaacggccatacat\n")
|
|
|
107 f.write(">Seq2\n")
|
|
|
108 f.write("agatgcgtaacggccataca\n")
|
|
|
109 f.write("gatgcgtaacggccatacat\n")
|
|
|
110
|
|
|
111 def _writeExpMapMultiFasta(self, mapFileName):
|
|
|
112 with open(mapFileName, 'w') as f:
|
|
|
113 f.write("caagatgcgtaacggccata_17\tSeq1\t4379\t4399\t17\n")
|
|
|
114 f.write("aagatgcgtaacggccatac_17\tSeq1\t4380\t4400\t17\n")
|
|
|
115 f.write("agatgcgtaacggccataca_16\tSeq1\t4381\t4401\t16\n")
|
|
|
116 f.write("gatgcgtaacggccatacat_16\tSeq1\t4382\t4402\t16\n")
|
|
|
117 f.write("agatgcgtaacggccataca_16\tSeq2\t4381\t4401\t16\n")
|
|
|
118 f.write("gatgcgtaacggccatacat_35\tSeq2\t4382\t4402\t35\n")
|
|
|
119
|
|
|
120 if __name__ == "__main__":
|
|
|
121 unittest.main() |
