-
Notifications
You must be signed in to change notification settings - Fork 6
/
benchmark.py
312 lines (241 loc) · 7.63 KB
/
benchmark.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import pyalign.problems
import pyalign.gaps
import pyalign.solve
import time
import collections
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import json
import Bio.Align
import parasail
from tqdm import tqdm
from pathlib import Path
from typing import Iterator
codomain_names = {
str(pyalign.solve.Score): "score",
str(pyalign.solve.Alignment): "alignment",
str(pyalign.solve.Solution): "matrix",
str(Iterator[pyalign.solve.Alignment]): "all alignments",
str(Iterator[pyalign.solve.Solution]): "all matrices"
}
ALPHABET = "ACGT"
class Aligner:
def prepare(self, a, b):
raise NotImplementedError()
def solve(self):
raise NotImplementedError()
@property
def name(self):
raise NotImplementedError()
@property
def num_problems(self):
raise NotImplementedError()
class PyAlignImplementation(Aligner):
def __init__(self, codomain=pyalign.solve.Alignment, encoded=False, batch=False):
self._codomain = codomain
self._encoded = encoded
self._batch = batch
self._solver = None
self._problem = None
self._num_problems = None
def prepare(self, a, b):
if not self._encoded:
pf = pyalign.problems.general(
pyalign.problems.Equality(eq=1, ne=0))
else:
pf = pyalign.problems.alphabetic(
ALPHABET,
pyalign.problems.Equality(eq=1, ne=0))
self._solver = pyalign.solve.LocalSolver(
gap_cost=pyalign.gaps.LinearGapCost(1),
codomain=self._codomain)
if not self._batch:
self._problem = pf.new_problem(a, b)
self._num_problems = 1
else:
self._num_problems = self._solver.batch_size
self._problem = [
pf.new_problem(a, b) for _ in range(self._num_problems)]
def solve(self):
return self._solver.solve(self._problem)
@property
def num_problems(self):
return self._num_problems
@property
def name(self):
terms = ["pyalign"]
if self._encoded:
terms.append("alphabet")
if self._batch:
batch_size = self._solver.batch_size
if batch_size > 1:
terms.append(f"SIMD-{batch_size * 32}")
else:
raise RuntimeError(
f"unexpected batch_size {batch_size}; running on generic build?")
return " +".join(terms)
class PurePythonImplementation(Aligner):
def __init__(self, backtrace=True):
self._aEncoded = None
self._bEncoded = None
self._aligner = None
self._v = None
self._backtrace = backtrace
def prepare(self, a, b):
# see https://github.com/eseraygun/python-alignment
from alignment.sequence import Sequence
from alignment.vocabulary import Vocabulary
from alignment.sequencealigner import SimpleScoring, LocalSequenceAligner
# Create sequences to be aligned.
a = Sequence(a)
b = Sequence(b)
# Create a vocabulary and encode the sequences.
v = Vocabulary()
self._v = v
self._aEncoded = v.encodeSequence(a)
self._bEncoded = v.encodeSequence(b)
# Create a scoring and align the sequences using global aligner.
scoring = SimpleScoring(1, 0)
self._aligner = LocalSequenceAligner(scoring, -1)
def solve(self):
# returns: score, encodeds
return self._aligner.align(
self._aEncoded, self._bEncoded, backtrace=self._backtrace)
def print(self, score, encodeds):
# Iterate over optimal alignments and print them.
for encoded in encodeds:
alignment = self._v.decodeSequenceAlignment(encoded)
@property
def name(self):
return "pure python"
@property
def num_problems(self):
return 1
class Pairwise2:
def __init__(self, **kwargs):
self._kwargs = kwargs
def prepare(self, a, b):
self._a = a
self._b = b
def solve(self):
aligner = Bio.Align.PairwiseAligner()
aligner.mode = 'local'
aligner.match_score = 1
aligner.mismatch_score = 0
aligner.open_gap_score = -1
aligner.extend_gap_score = -1
aligner.align(self._a, self._b)
@property
def name(self):
return "Bio.Align.PairwiseAligner"
@property
def num_problems(self):
return 1
class Parasail:
def __init__(self, **kwargs):
self._kwargs = kwargs
self._matrix = parasail.matrix_create("ACGT", 1, 0)
def prepare(self, a, b):
self._a = a
self._b = b
def solve(self):
parasail.sw(self._a, self._b, 1, 1, self._matrix)
@property
def name(self):
return "parasail.sw"
@property
def num_problems(self):
return 1
def benchmark(num_runs=1000, seq_lens=(20, 20), is_large_seq=False):
seq_gen = pyalign.utils.RandomSequenceGenerator(ALPHABET, seed=24242)
a = seq_gen(seq_lens[0])
b = seq_gen(seq_lens[1])
codomains = [
pyalign.solve.Score,
pyalign.solve.Alignment
]
if not is_large_seq:
codomains.extend([
pyalign.solve.Solution,
Iterator[pyalign.solve.Alignment],
Iterator[pyalign.solve.Solution]
])
def aligners():
if not is_large_seq:
yield str(pyalign.solve.Score), PurePythonImplementation(backtrace=False)
yield str(pyalign.solve.Alignment), PurePythonImplementation(backtrace=True)
yield str(pyalign.solve.Score), Parasail()
yield str(pyalign.solve.Alignment), Parasail()
yield str(pyalign.solve.Score), Pairwise2(score_only=True)
yield str(pyalign.solve.Alignment), Pairwise2(one_alignment_only=True)
for batch in ((False,) if is_large_seq else (False, True)):
for encoded in ((True,) if is_large_seq else (False, True)):
for codomain in codomains:
yield str(codomain), PyAlignImplementation(
codomain, encoded=encoded, batch=batch)
path = Path(f"runtimes_{seq_lens[0]}_{seq_lens[1]}.json")
if path.exists():
with open(path, "r") as f:
runtimes_μs = json.loads(f.read())
else:
runtimes_μs = collections.defaultdict(dict)
μs_to_ns = 1000
for codomain, aligner in tqdm(list(aligners())):
aligner.prepare(a, b)
t0 = time.perf_counter_ns()
for _ in range(num_runs):
aligner.solve()
t1 = time.perf_counter_ns()
runtimes_μs[codomain][aligner.name] = (t1 - t0) // (μs_to_ns * num_runs * aligner.num_problems)
with open(path, "w") as f:
f.write(json.dumps(runtimes_μs))
def variant_sort_order(s):
if s == "pure python":
return ""
else:
return s
variants = set()
for codomain, times in runtimes_μs.items():
for k in times.keys():
variants.add(k)
variants = sorted(list(variants), key=variant_sort_order)
y = dict()
for variant in variants:
ys = []
for codomain in codomains:
ys.append(runtimes_μs[str(codomain)].get(variant, np.nan))
y[variant] = np.array(ys, dtype=np.float32)
y_median = np.nanmedian(np.concatenate(list(y.values())))
if y_median > 1000:
time_unit = 'ms'
y = dict((k, v / 1000) for k, v in y.items())
else:
time_unit = 'μs'
x = np.arange(0, len(codomains) * len(variants), len(variants)) # the label locations
width = 0.6
x_c = x - ((len(variants) - 1) / 2) * width
cmap = matplotlib.cm.get_cmap('Set2')
norm = matplotlib.colors.Normalize(vmin=0, vmax=len(variants) - 1)
fig, ax = plt.subplots(figsize=(12, 6))
for i, variant in enumerate(variants):
ax.bar(x_c + width * i, y[variant], width, label=variant, color=cmap(norm(i)))
ax.set_ylabel(f'time in {time_unit}')
#ax.set_yscale('log')
from matplotlib.ticker import StrMethodFormatter, LogLocator
ax.yaxis.set_major_formatter(StrMethodFormatter('{x:.0f}'))
ax.yaxis.set_minor_formatter(StrMethodFormatter('{x:.0f}'))
#ax.yaxis.set_major_locator(LogLocator(subs=(0.75,)))
ax.set_xticks(x)
ax.set_xticklabels([codomain_names[str(s)] for s in codomains])
ax.legend(loc="center right", bbox_to_anchor=(1.3, 0.9))
#plt.yticks(rotation=45)
plt.grid(which="both", alpha=0.25)
plt.title(f"local alignment, linear gap cost\nsequence lengths {seq_lens[0]} and {seq_lens[1]}")
fig.tight_layout()
plt.savefig(f'benchmark_{seq_lens[0]}_{seq_lens[1]}.svg', bbox_inches='tight')
if __name__ == "__main__":
benchmark(seq_lens=(10, 100), is_large_seq=False)
benchmark(seq_lens=(5000, 10000), is_large_seq=True, num_runs=10)