FunVIP 0.3.20__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (36) hide show
  1. FunVIP-0.3.20.dist-info/LICENSE +674 -0
  2. FunVIP-0.3.20.dist-info/METADATA +32 -0
  3. FunVIP-0.3.20.dist-info/RECORD +36 -0
  4. FunVIP-0.3.20.dist-info/WHEEL +5 -0
  5. FunVIP-0.3.20.dist-info/entry_points.txt +3 -0
  6. FunVIP-0.3.20.dist-info/top_level.txt +3 -0
  7. data/__init__.py +0 -0
  8. external/BLAST_Windows/bin/cleanup-blastdb-volumes.py +162 -0
  9. external/__init__.py +0 -0
  10. src/__init__.py +0 -0
  11. src/align.py +124 -0
  12. src/cluster.py +510 -0
  13. src/command.py +360 -0
  14. src/concatenate.py +356 -0
  15. src/dataset.py +716 -0
  16. src/ext.py +448 -0
  17. src/hasher.py +98 -0
  18. src/initialize.py +335 -0
  19. src/logger.py +69 -0
  20. src/logics.py +104 -0
  21. src/modeltest.py +443 -0
  22. src/ncbi.py +160 -0
  23. src/opt_generator.py +72 -0
  24. src/patch.py +261 -0
  25. src/reporter.py +875 -0
  26. src/save.py +181 -0
  27. src/search.py +440 -0
  28. src/tool.py +309 -0
  29. src/tree.py +222 -0
  30. src/tree_interpretation.py +1379 -0
  31. src/tree_interpretation_pipe.py +679 -0
  32. src/trim.py +118 -0
  33. src/validate_input.py +846 -0
  34. src/validate_option.py +1609 -0
  35. src/validation.py +38 -0
  36. src/version.py +337 -0
src/validation.py ADDED
@@ -0,0 +1,38 @@
1
+ from Bio import SeqIO
2
+ from Bio.Seq import Seq
3
+ from Bio.SeqRecord import SeqRecord
4
+
5
+ from io import StringIO
6
+
7
+ # from .logger import Mes
8
+
9
+
10
+ # validate if string is good sequence
11
+ def validate_seq(seqstring):
12
+ if seqstring.startswith(">"):
13
+ tmp = StringIO(seqstring)
14
+
15
+ try:
16
+ seqlist = list(SeqIO.parse(tmp, fasta))
17
+ return "seqrecord", seqlist
18
+ except:
19
+ logging.warning(f"Invalid seqrecord {seqstring}")
20
+ return "invalid", None
21
+
22
+ else:
23
+ seqstring = seqstring.replace(" ", "").replace("\n", "")
24
+
25
+ try:
26
+ seq = Seq(seqstring)
27
+ seqlist = [
28
+ SeqRecord(
29
+ seq,
30
+ id="input",
31
+ description="tmp",
32
+ annotations={"molecule_type": "DNA"},
33
+ )
34
+ ]
35
+ return "seqrecord", seqlist
36
+ except:
37
+ logging.warning(f"Invalid sequence {seqstring}")
38
+ return "invalid", None
src/version.py ADDED
@@ -0,0 +1,337 @@
1
+ # Version management module
2
+ import sys
3
+ import subprocess
4
+ from importlib.metadata import version
5
+
6
+ # import
7
+
8
+
9
+ """
10
+ version = {
11
+ "FunVIP": "0.3.19.0.1.3",
12
+ "BLASTn": "",
13
+ "MMseqs2": "",
14
+ "MAFFT": "",
15
+ "TrimAl": "",
16
+ "Gblocks": "0.91b",
17
+ "FastTree": "",
18
+ "IQTREE2": "",
19
+ "RAxML": "",
20
+ }
21
+ """
22
+
23
+
24
+ class Version:
25
+ def __init__(self, opt, path):
26
+ self.FunVIP = ""
27
+ self.GenMine = ""
28
+ self.BLASTn = ""
29
+ self.MMseqs2 = ""
30
+ self.MAFFT = ""
31
+ self.trimAl = ""
32
+ self.Gblocks = "0.91b"
33
+ self.Modeltest_NG = ""
34
+ self.FastTree = ""
35
+ self.IQTREE2 = ""
36
+ self.RAxML = ""
37
+
38
+ # For windows platform
39
+ if sys.platform == "win32":
40
+ ### FunVIP
41
+ self.FunVIP = version("FunVIP")
42
+
43
+ ### GenMine
44
+ self.GenMine = version("GenMine")
45
+
46
+ ### BLASTn
47
+ CMD = [f"{path.sys_path}/external/BLAST_Windows/bin/blastn.exe", "-version"]
48
+ result = subprocess.Popen(
49
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
50
+ )
51
+ stdout, stderr = result.communicate()
52
+ stdout_str = stdout.decode("utf-8")
53
+ ## tableformat
54
+ # blastn: blastn: 2.12.0+
55
+ # Package: blast 2.12.0, build Jun 4 2021 03:25:07
56
+ self.BLASTn = stdout_str.split("\n")[0].split(" ")[1].strip()
57
+ # print("BLASTn", self.BLASTn)
58
+
59
+ ### MMSeqs2
60
+ CMD = [
61
+ f"{path.sys_path}/external/mmseqs_Windows/mmseqs.bat",
62
+ "-h",
63
+ ]
64
+ result = subprocess.Popen(
65
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
66
+ )
67
+ stdout, stderr = result.communicate()
68
+ stdout_str = stdout.decode("utf-8")
69
+ self.MMseqs2 = stdout_str.split("Version: ")[1].split("\n")[0].strip()
70
+ # print("MMseqs2", self.MMseqs2)
71
+
72
+ ### MAFFT
73
+ CMD = [
74
+ f"{path.sys_path}/external/MAFFT_Windows/mafft-win/mafft.bat",
75
+ "--version",
76
+ ]
77
+ result = subprocess.Popen(
78
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
79
+ )
80
+ stdout, stderr = result.communicate()
81
+ # MAFFT, output is on the stderr
82
+ stderr_str = stderr.decode("utf-8")
83
+ self.MAFFT = stderr_str.split("\n")[-2].split(" ")[0].strip()
84
+ # print("MAFFT", self.MAFFT)
85
+
86
+ ### TrimAl
87
+ CMD = [
88
+ f"{path.sys_path}/external/trimal.v1.4/trimAl/bin/trimal.exe",
89
+ "--version",
90
+ ]
91
+ result = subprocess.Popen(
92
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
93
+ )
94
+ stdout, stderr = result.communicate()
95
+ stdout_str = stdout.decode("utf-8")
96
+ self.trimAl = stdout_str.split("\n")[1].split(" ")[1]
97
+ # print("trimAl", self.trimAl)
98
+
99
+ ### Modeltest-ng
100
+ ## Not supported in Windows
101
+ self.Modeltest_NG = "not supported"
102
+
103
+ ### FastTree
104
+ ## Also use stderr of FastTree
105
+ CMD = [
106
+ f"{path.sys_path}/external/FastTree_Windows/FastTree.exe",
107
+ "-expert",
108
+ ]
109
+
110
+ result = subprocess.Popen(
111
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
112
+ )
113
+ stdout, stderr = result.communicate()
114
+ stdout_str = stdout.decode("utf-8")
115
+ stderr_str = stderr.decode("utf-8")
116
+ self.FastTree = stderr_str.split(" ")[4]
117
+ # print("FastTree", self.FastTree)
118
+
119
+ ### IQTREE
120
+ CMD = [
121
+ f"{path.sys_path}/external/iqtree/bin/iqtree2.exe",
122
+ "--version",
123
+ ]
124
+ result = subprocess.Popen(
125
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
126
+ )
127
+ stdout, stderr = result.communicate()
128
+ stdout_str = stdout.decode("utf-8")
129
+ self.IQTREE2 = stdout_str.split(" ")[3]
130
+ # print("IQTREE2", self.IQTREE2)
131
+
132
+ ### RAxML
133
+ if opt.avx is True:
134
+ CMD = [
135
+ f"{path.sys_path}/external/RAxML_Windows/raxmlHPC-PTHREADS-AVX2.exe",
136
+ "-v",
137
+ ]
138
+ else:
139
+ CMD = [
140
+ f"{path.sys_path}/external/RAxML_Windows/raxmlHPC-PTHREADS-SSE3.exe",
141
+ "-v",
142
+ ]
143
+
144
+ result = subprocess.Popen(
145
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
146
+ )
147
+ stdout, stderr = result.communicate()
148
+ stdout_str = stdout.decode("utf-8")
149
+ self.RAxML = stdout_str.split("\n")[2].split(" ")[4]
150
+ # print("RAxML", self.RAxML)
151
+
152
+ # For apple silicon platform
153
+ elif sys.platform == "darwin":
154
+ ### FunVIP
155
+ self.FunVIP = version("FunVIP")
156
+
157
+ ### GenMine
158
+ self.GenMine = version("GenMine")
159
+
160
+ ### BLASTn
161
+ CMD = ["blastn", "-version"]
162
+ result = subprocess.Popen(
163
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
164
+ )
165
+ stdout, stderr = result.communicate()
166
+ stdout_str = stdout.decode("utf-8")
167
+ ## Format
168
+ # blastn: blastn: 2.12.0+
169
+ # Package: blast 2.12.0, build Jun 4 2021 03:25:07
170
+ self.BLASTn = stdout_str.split("\n")[0].split(" ")[1].strip()
171
+ # print("BLASTn", self.BLASTn)
172
+
173
+ ### MMSeqs2
174
+ CMD = ["mmseqs", "-h"]
175
+ result = subprocess.Popen(
176
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
177
+ )
178
+ stdout, stderr = result.communicate()
179
+ stdout_str = stdout.decode("utf-8")
180
+ self.MMseqs2 = stdout_str.split("Version: ")[1].split("\n")[0].strip()
181
+ # print("MMseqs2", self.MMseqs2)
182
+
183
+ ### MAFFT
184
+ CMD = ["mafft", "--version"]
185
+ result = subprocess.Popen(
186
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
187
+ )
188
+ stdout, stderr = result.communicate()
189
+ # MAFFT, output is on the stderr
190
+ stderr_str = stderr.decode("utf-8")
191
+ self.MAFFT = stderr_str.split("\n")[-2].split(" ")[0].strip()
192
+ # print("MAFFT", self.MAFFT)
193
+
194
+ ### TrimAl
195
+ CMD = ["trimal", "--version"]
196
+ result = subprocess.Popen(
197
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
198
+ )
199
+ stdout, stderr = result.communicate()
200
+ stdout_str = stdout.decode("utf-8")
201
+ self.trimAl = stdout_str.split("\n")[1].split(" ")[1]
202
+ # print("trimAl", self.trimAl)
203
+
204
+ ### Modeltest-ng
205
+ ## Not supported in apple silicon
206
+ self.Modeltest_NG = "not supported"
207
+
208
+ ### FastTree
209
+ ## Also use stderr of FastTree
210
+ CMD = ["FastTree", "-expert"]
211
+ result = subprocess.Popen(
212
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
213
+ )
214
+ stdout, stderr = result.communicate()
215
+ stderr_str = stderr.decode("utf-8")
216
+ self.FastTree = stderr_str.split(" ")[4]
217
+ # print("FastTree", self.FastTree)
218
+
219
+ ### IQTREE
220
+ CMD = ["iqtree", "--version"]
221
+ result = subprocess.Popen(
222
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
223
+ )
224
+ stdout, stderr = result.communicate()
225
+ stdout_str = stdout.decode("utf-8")
226
+ self.IQTREE2 = stdout_str.split(" ")[3]
227
+ # print("IQTREE2", self.IQTREE2)
228
+
229
+ ### RAxML
230
+ if opt.avx is True:
231
+ CMD = ["raxmlHPC-PTHREADS-AVX2", "-v"]
232
+ else:
233
+ CMD = ["raxmlHPC-PTHREADS-SSE3", "-v"]
234
+
235
+ result = subprocess.Popen(
236
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
237
+ )
238
+ stdout, stderr = result.communicate()
239
+ stdout_str = stdout.decode("utf-8")
240
+ self.RAxML = stdout_str.split("\n")[2].split(" ")[4]
241
+ # print("RAxML", self.RAxML)
242
+
243
+ # For linux platform
244
+ else:
245
+ ### FunVIP
246
+ self.FunVIP = version("FunVIP")
247
+
248
+ ### GenMine
249
+ self.GenMine = version("GenMine")
250
+
251
+ ### BLASTn
252
+ CMD = ["blastn", "-version"]
253
+ result = subprocess.Popen(
254
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
255
+ )
256
+ stdout, stderr = result.communicate()
257
+ stdout_str = stdout.decode("utf-8")
258
+ ## Format
259
+ # blastn: blastn: 2.12.0+
260
+ # Package: blast 2.12.0, build Jun 4 2021 03:25:07
261
+ self.BLASTn = stdout_str.split("\n")[0].split(" ")[1].strip()
262
+ # print("BLASTn", self.BLASTn)
263
+
264
+ ### MMSeqs2
265
+ CMD = ["mmseqs", "-h"]
266
+ result = subprocess.Popen(
267
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
268
+ )
269
+ stdout, stderr = result.communicate()
270
+ stdout_str = stdout.decode("utf-8")
271
+ self.MMseqs2 = stdout_str.split("Version: ")[1].split("\n")[0].strip()
272
+ # print("MMseqs2", self.MMseqs2)
273
+
274
+ ### MAFFT
275
+ CMD = ["mafft", "--version"]
276
+ result = subprocess.Popen(
277
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
278
+ )
279
+ stdout, stderr = result.communicate()
280
+ # MAFFT, output is on the stderr
281
+ stderr_str = stderr.decode("utf-8")
282
+ self.MAFFT = stderr_str.split("\n")[-2].split(" ")[0].strip()
283
+ # print("MAFFT", self.MAFFT)
284
+
285
+ ### TrimAl
286
+ CMD = ["trimal", "--version"]
287
+ result = subprocess.Popen(
288
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
289
+ )
290
+ stdout, stderr = result.communicate()
291
+ stdout_str = stdout.decode("utf-8")
292
+ self.trimAl = stdout_str.split("\n")[1].split(" ")[1]
293
+ # print("trimAl", self.trimAl)
294
+
295
+ ### Modeltest-ng
296
+ ## Not supported in Windows
297
+ CMD = ["modeltest-ng", "--version"]
298
+ result = subprocess.Popen(
299
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
300
+ )
301
+ stdout, stderr = result.communicate()
302
+ stdout_str = stdout.decode("utf-8")
303
+ self.Modeltest_NG = stdout_str.split("ModelTest-NG ")[1].split(" ")[0]
304
+
305
+ ### FastTree
306
+ ## Also use stderr of FastTree
307
+ CMD = ["FastTree", "-expert"]
308
+ result = subprocess.Popen(
309
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
310
+ )
311
+ stdout, stderr = result.communicate()
312
+ stderr_str = stderr.decode("utf-8")
313
+ self.FastTree = stderr_str.split(" ")[4]
314
+ # print("FastTree", self.FastTree)
315
+
316
+ ### IQTREE
317
+ CMD = ["iqtree", "--version"]
318
+ result = subprocess.Popen(
319
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
320
+ )
321
+ stdout, stderr = result.communicate()
322
+ stdout_str = stdout.decode("utf-8")
323
+ self.IQTREE2 = stdout_str.split(" ")[3]
324
+ # print("IQTREE2", self.IQTREE2)
325
+
326
+ ### RAxML
327
+ if opt.avx is True:
328
+ CMD = ["raxmlHPC-PTHREADS-AVX2", "-v"]
329
+ else:
330
+ CMD = ["raxmlHPC-PTHREADS-SSE3", "-v"]
331
+
332
+ result = subprocess.Popen(
333
+ CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE
334
+ )
335
+ stdout, stderr = result.communicate()
336
+ stdout_str = stdout.decode("utf-8")
337
+ self.RAxML = stdout_str.split("\n")[2].split(" ")[4]