|
1
|
|
|
import diff_classifier.features as ft |
|
2
|
|
|
import diff_classifier.msd as msd |
|
3
|
|
|
import numpy.testing as npt |
|
4
|
|
|
import pandas.util.testing as pdt |
|
5
|
|
|
import numpy as np |
|
6
|
|
|
import pandas as pd |
|
7
|
|
|
import math |
|
8
|
|
|
|
|
9
|
|
|
|
|
10
|
|
|
def test_make_xyarray(): |
|
11
|
|
|
d = {'Frame': [0, 1, 2, 3, 4, 0, 1, 2, 3, 4], |
|
12
|
|
|
'Track_ID': [1, 1, 1, 1, 1, 2, 2, 2, 2, 2], |
|
13
|
|
|
'X': [np.nan, 6, 7, 8, 9, 1, 2, 3, 4, np.nan], |
|
14
|
|
|
'Y': [np.nan, 7, 8, 9, 10, 2, 3, 4, 5, np.nan]} |
|
15
|
|
|
df = pd.DataFrame(data=d) |
|
16
|
|
|
cols = ['Frame', 'Track_ID', 'X', 'Y', 'MSDs', 'Gauss'] |
|
17
|
|
|
length = max(df['Frame']) + 1 |
|
18
|
|
|
m_df = msd.all_msds2(df, frames=length)[cols] |
|
19
|
|
|
|
|
20
|
|
|
dt = {'Frame': [float(i) for i in[0, 1, 2, 3]], |
|
21
|
|
|
'Track_ID': [float(i) for i in[2, 2, 2, 2]], |
|
22
|
|
|
'X': [float(i) for i in[1, 2, 3, 4]], |
|
23
|
|
|
'Y': [float(i) for i in[2, 3, 4, 5]], |
|
24
|
|
|
'MSDs': [float(i) for i in[0, 2, 8, 18]], |
|
25
|
|
|
'Gauss': [float(i) for i in[0, 0.25, 0.25, 0.25]]} |
|
26
|
|
|
dft = pd.DataFrame(data=dt) |
|
27
|
|
|
|
|
28
|
|
|
pdt.assert_frame_equal(ft.unmask_track(m_df[m_df['Track_ID']==2]), dft) |
|
29
|
|
|
|
|
30
|
|
View Code Duplication |
def test_alpha_calc(): |
|
|
|
|
|
|
31
|
|
|
frames = 5 |
|
32
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
33
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
34
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
35
|
|
|
'Track_ID': np.ones(frames)} |
|
36
|
|
|
df = pd.DataFrame(data=d) |
|
37
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
38
|
|
|
assert ft.alpha_calc(df) == (2.0000000000000004, 0.4999999999999998) |
|
39
|
|
|
|
|
40
|
|
|
frames = 10 |
|
41
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
42
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+5), |
|
43
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
44
|
|
|
'Track_ID': np.ones(frames)} |
|
45
|
|
|
df = pd.DataFrame(data=d) |
|
46
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
47
|
|
|
assert ft.alpha_calc(df) == (0.8201034110620524, 0.1494342948594476) |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
def test_gyration_tensor(): |
|
51
|
|
|
frames = 6 |
|
52
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
53
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
54
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
55
|
|
|
'Track_ID': np.ones(frames)} |
|
56
|
|
|
df = pd.DataFrame(data=d) |
|
57
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
58
|
|
|
o1, o2, o3, o4 = (8.0, 0.0, np.array([ 0.70710678, -0.70710678]), np.array([0.70710678, 0.70710678])) |
|
59
|
|
|
d1, d2, d3, d4 = ft.gyration_tensor(df) |
|
60
|
|
|
|
|
61
|
|
|
assert d1 == o1 |
|
62
|
|
|
assert d2 == o2 |
|
63
|
|
|
npt.assert_almost_equal(o3, d3) |
|
64
|
|
|
npt.assert_almost_equal(o4, d4) |
|
65
|
|
|
|
|
66
|
|
|
frames = 10 |
|
67
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
68
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+5), |
|
69
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+5), |
|
70
|
|
|
'Track_ID': np.ones(frames)} |
|
71
|
|
|
df = pd.DataFrame(data=d) |
|
72
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
73
|
|
|
o1, o2, o3, o4 = (0.47248734315843355, 0.3447097846562249, np.array([0.83907153, 0.54402111]), |
|
74
|
|
|
np.array([-0.54402111, 0.83907153])) |
|
75
|
|
|
d1, d2, d3, d4 = ft.gyration_tensor(df) |
|
76
|
|
|
|
|
77
|
|
|
assert d1 == o1 |
|
78
|
|
|
assert d2 == o2 |
|
79
|
|
|
npt.assert_almost_equal(o3, d3) |
|
80
|
|
|
npt.assert_almost_equal(o4, d4) |
|
81
|
|
|
|
|
82
|
|
|
|
|
83
|
|
View Code Duplication |
def test_kurtosis(): |
|
|
|
|
|
|
84
|
|
|
frames = 5 |
|
85
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
86
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
87
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
88
|
|
|
'Track_ID': np.ones(frames)} |
|
89
|
|
|
df = pd.DataFrame(data=d) |
|
90
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
91
|
|
|
assert ft.kurtosis(df) == 4.079999999999999 |
|
92
|
|
|
|
|
93
|
|
|
frames = 10 |
|
94
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
95
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+3), |
|
96
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
97
|
|
|
'Track_ID': np.ones(frames)} |
|
98
|
|
|
df = pd.DataFrame(data=d) |
|
99
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
100
|
|
|
assert ft.kurtosis(df) == 1.4759027695843443 |
|
101
|
|
|
|
|
102
|
|
|
|
|
103
|
|
|
def test_asymmetry(): |
|
104
|
|
|
frames = 10 |
|
105
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
106
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
107
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
108
|
|
|
'Track_ID': np.ones(frames)} |
|
109
|
|
|
df = pd.DataFrame(data=d) |
|
110
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
111
|
|
|
|
|
112
|
|
|
o1, o2, o3, o4, o5 = (20.0, 0.0, 1.0, 0.0, 0.69314718) |
|
113
|
|
|
d1, d2, d3, d4, d5 = ft.asymmetry(df) |
|
114
|
|
|
assert math.isclose(o1, d1, abs_tol=1e-10) |
|
115
|
|
|
assert math.isclose(o2, d2, abs_tol=1e-10) |
|
116
|
|
|
assert math.isclose(o3, d3, abs_tol=1e-10) |
|
117
|
|
|
assert math.isclose(o4, d4, abs_tol=1e-10) |
|
118
|
|
|
assert math.isclose(o5, d5, abs_tol=1e-10) |
|
119
|
|
|
|
|
120
|
|
|
frames = 100 |
|
121
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
122
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+3), |
|
123
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
124
|
|
|
'Track_ID': np.ones(frames)} |
|
125
|
|
|
df = pd.DataFrame(data=d) |
|
126
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
127
|
|
|
|
|
128
|
|
|
o1, o2, o3, o4, o5 = (0.4254120816156, 0.42004967815488, 0.0001609000151811, 0.9873948021401, 2.0114322402896e-05) |
|
129
|
|
|
d1, d2, d3, d4, d5 = ft.asymmetry(df) |
|
130
|
|
|
assert math.isclose(o1, d1) |
|
131
|
|
|
assert math.isclose(o2, d2) |
|
132
|
|
|
assert math.isclose(o3, d3) |
|
133
|
|
|
assert math.isclose(o4, d4) |
|
134
|
|
|
assert math.isclose(o5, d5) |
|
135
|
|
|
|
|
136
|
|
|
|
|
137
|
|
|
def test_minBoundingRect(): |
|
138
|
|
|
frames = 10 |
|
139
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
140
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
141
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
142
|
|
|
'Track_ID': np.ones(frames)} |
|
143
|
|
|
df = pd.DataFrame(data=d) |
|
144
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
145
|
|
|
|
|
146
|
|
|
d1, d2, d3, d4, d5, d6 = ft.minBoundingRect(df) |
|
147
|
|
|
o1, o2, o3, o4 = (-2.356194490192, 0, 14.142135623730, 0) |
|
148
|
|
|
o5 = np.array([10, 8]) |
|
149
|
|
|
o6 = np.array([[5., 3.], [15., 13.], [15., 13.], [5., 3.]]) |
|
150
|
|
|
|
|
151
|
|
|
#assert math.isclose(d1, o1, abs_tol=1e-10) |
|
152
|
|
|
assert math.isclose(d2, o2, abs_tol=1e-10) |
|
153
|
|
|
assert math.isclose(d3, o3, abs_tol=1e-10) |
|
154
|
|
|
assert math.isclose(d4, o4, abs_tol=1e-10) |
|
155
|
|
|
npt.assert_almost_equal(d5, o5) |
|
156
|
|
|
#npt.assert_almost_equal(d6, o6) |
|
157
|
|
|
|
|
158
|
|
|
frames = 100 |
|
159
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
160
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+3), |
|
161
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
162
|
|
|
'Track_ID': np.ones(frames)} |
|
163
|
|
|
df = pd.DataFrame(data=d) |
|
164
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
165
|
|
|
|
|
166
|
|
|
d1, d2, d3, d4, d5, d6 = ft.minBoundingRect(df) |
|
167
|
|
|
o1, o2, o3, o4 = (-2.7345175425633, 3.7067697307443, 1.899593160348, 1.951349272106) |
|
168
|
|
|
o5 = np.array([-0.00098312, 0.00228019]) |
|
169
|
|
|
o6 = np.array([[-1.2594591, 0.52217706], |
|
170
|
|
|
[0.4849046, 1.27427376], |
|
171
|
|
|
[1.25749286, -0.51761668], |
|
172
|
|
|
[-0.48687084, -1.26971339]]) |
|
173
|
|
|
|
|
174
|
|
|
#assert math.isclose(d1, o1, abs_tol=1e-10) |
|
175
|
|
|
assert math.isclose(d2, o2, abs_tol=1e-10) |
|
176
|
|
|
assert math.isclose(d3, o3, abs_tol=1e-10) |
|
177
|
|
|
assert math.isclose(d4, o4, abs_tol=1e-10) |
|
178
|
|
|
npt.assert_almost_equal(d5, o5) |
|
179
|
|
|
#npt.assert_almost_equal(d6, o6) |
|
180
|
|
|
|
|
181
|
|
|
def test_aspectratio(): |
|
182
|
|
|
frames = 6 |
|
183
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
184
|
|
|
'X': [0, 1, 1, 2, 2, 3], |
|
185
|
|
|
'Y': [0, 0, 1, 1, 2, 2], |
|
186
|
|
|
'Track_ID': np.ones(frames)} |
|
187
|
|
|
df = pd.DataFrame(data=d) |
|
188
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
189
|
|
|
assert ft.aspectratio(df)[0:2] == (3.9000000000000026, 0.7435897435897438) |
|
190
|
|
|
npt.assert_almost_equal(ft.aspectratio(df)[2], np.array([1.5, 1. ])) |
|
191
|
|
|
|
|
192
|
|
|
|
|
193
|
|
View Code Duplication |
def test_boundedness(): |
|
|
|
|
|
|
194
|
|
|
frames = 100 |
|
195
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
196
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+3), |
|
197
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
198
|
|
|
'Track_ID': np.ones(frames)} |
|
199
|
|
|
df = pd.DataFrame(data=d) |
|
200
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
201
|
|
|
assert ft.boundedness(df) == (0.607673328076712, 5.674370543833708, -0.0535555587618044) |
|
202
|
|
|
|
|
203
|
|
|
frames = 10 |
|
204
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
205
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
206
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
207
|
|
|
'Track_ID': np.ones(frames)} |
|
208
|
|
|
df = pd.DataFrame(data=d) |
|
209
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
210
|
|
|
assert ft.boundedness(df) == (0.039999999999999994, 1.0, -0.21501108474766228) |
|
211
|
|
|
|
|
212
|
|
|
|
|
213
|
|
View Code Duplication |
def test_efficiency(): |
|
|
|
|
|
|
214
|
|
|
frames = 100 |
|
215
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
216
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+3), |
|
217
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
218
|
|
|
'Track_ID': np.ones(frames)} |
|
219
|
|
|
df = pd.DataFrame(data=d) |
|
220
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
221
|
|
|
|
|
222
|
|
|
assert ft.efficiency(df) == (0.003548421265914009, 0.0059620286331768385) |
|
223
|
|
|
|
|
224
|
|
|
frames = 10 |
|
225
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
226
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
227
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
228
|
|
|
'Track_ID': np.ones(frames)} |
|
229
|
|
|
df = pd.DataFrame(data=d) |
|
230
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
231
|
|
|
|
|
232
|
|
|
assert ft.efficiency(df) == (10.0, 1.0) |
|
233
|
|
|
|
|
234
|
|
|
def test_msd_ratio(): |
|
235
|
|
|
frames = 10 |
|
236
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
237
|
|
|
'X': np.sin(np.linspace(0, frames, frames)+3), |
|
238
|
|
|
'Y': np.cos(np.linspace(0, frames, frames)+3), |
|
239
|
|
|
'Track_ID': np.ones(frames)} |
|
240
|
|
|
df = pd.DataFrame(data=d) |
|
241
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
242
|
|
|
|
|
243
|
|
|
assert ft.msd_ratio(df, 1, 9) == 0.09708430006771959 |
|
244
|
|
|
|
|
245
|
|
|
frames = 10 |
|
246
|
|
|
d = {'Frame': np.linspace(0, frames, frames), |
|
247
|
|
|
'X': np.linspace(0, frames, frames)+5, |
|
248
|
|
|
'Y': np.linspace(0, frames, frames)+3, |
|
249
|
|
|
'Track_ID': np.ones(frames)} |
|
250
|
|
|
df = pd.DataFrame(data=d) |
|
251
|
|
|
df = msd.all_msds2(df, frames=frames+1) |
|
252
|
|
|
|
|
253
|
|
|
assert ft.msd_ratio(df, 1, 9) == -0.09876543209876543 |
|
254
|
|
|
|
|
255
|
|
|
# def test_calculate_features(): |
|
256
|
|
|
# d = {'Frame': [0, 1, 2, 3, 4, 0, 1, 2, 3, 4], |
|
257
|
|
|
# 'Track_ID': [1, 1, 1, 1, 1, 2, 2, 2, 2, 2], |
|
258
|
|
|
# 'X': [0, 0, 1, 1, 2, 1, 1, 2, 2, 3], |
|
259
|
|
|
# 'Y': [0, 1, 1, 2, 2, 0, 1, 1, 2, 2]} |
|
260
|
|
|
# df = pd.DataFrame(data=d) |
|
261
|
|
|
# dfi = msd.all_msds2(df, frames = 5) |
|
262
|
|
|
# feat = ft.calculate_features(dfi) |
|
263
|
|
|
|
|
264
|
|
|
# d = {'AR': np.ones(2)*3.9999999999999996, |
|
265
|
|
|
# 'D_fit': np.ones(2)*0.1705189932550273, |
|
266
|
|
|
# 'MSD_ratio': np.ones(2)*-0.2666666666666666, |
|
267
|
|
|
# 'X': [0.75, 1.75], |
|
268
|
|
|
# 'Y': [1.25, 1.25], |
|
269
|
|
|
# 'Track_ID': [1.0, 2.0], |
|
270
|
|
|
# 'alpha': np.ones(2)*1.7793370720777268, |
|
271
|
|
|
# 'asymmetry1': np.ones(2)*0.9440237239896903, |
|
272
|
|
|
# 'asymmetry2': np.ones(2)*0.12, |
|
273
|
|
|
# 'asymmetry3': np.ones(2)*0.3691430189107616, |
|
274
|
|
|
# 'boundedness': np.ones(2)*0.25, |
|
275
|
|
|
# 'efficiency': np.ones(2)*2.0, |
|
276
|
|
|
# 'elongation': np.ones(2)*0.75, |
|
277
|
|
|
# 'fractal_dim': np.ones(2)*1.333333333333333, |
|
278
|
|
|
# 'frames': [5.0, 5.0], |
|
279
|
|
|
# 'kurtosis': np.ones(2)*1.166666666666667, |
|
280
|
|
|
# 'straightness': np.ones(2)*0.7071067811865476, |
|
281
|
|
|
# 'trappedness': np.ones(2)*-0.15258529289428524} |
|
282
|
|
|
# dfi = pd.DataFrame(data=d) |
|
283
|
|
|
|
|
284
|
|
|
# pdt.assert_frame_equal(dfi, feat) |
|
285
|
|
|
|
|
286
|
|
|
def test_unmask_track(): |
|
287
|
|
|
size = 10 |
|
288
|
|
|
ID = np.ones(size) |
|
289
|
|
|
frame = np.linspace(5, size-1+5, size) |
|
290
|
|
|
x = frame + 1 |
|
291
|
|
|
y = frame + 3 |
|
292
|
|
|
|
|
293
|
|
|
d = {'Frame': frame, |
|
294
|
|
|
'Track_ID': ID, |
|
295
|
|
|
'X': x, |
|
296
|
|
|
'Y': y} |
|
297
|
|
|
di = pd.DataFrame(data=d) |
|
298
|
|
|
track = msd.all_msds2(di, frames=20) |
|
299
|
|
|
output = ft.unmask_track(track) |
|
300
|
|
|
|
|
301
|
|
|
d2 = {'Frame': frame-5, |
|
302
|
|
|
'Track_ID': ID, |
|
303
|
|
|
'X': x, |
|
304
|
|
|
'Y': y, |
|
305
|
|
|
'MSDs': np.array((0, 2, 8, 18, 32, 50, 72, 98, 128, 162)).astype('float64'), |
|
306
|
|
|
'Gauss': np.array((0, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25, 0.25))} |
|
307
|
|
|
check = pd.DataFrame(data=d2) |
|
308
|
|
|
|
|
309
|
|
|
pdt.assert_frame_equal(output, check) |
ccurtis7 /
diff_classifier
Loading history...