|
1
|
|
|
import boto3 |
|
2
|
|
|
import cloudpickle |
|
3
|
|
|
import os |
|
4
|
|
|
import pickle |
|
5
|
|
|
from argparse import ArgumentParser |
|
6
|
|
|
from functools import wraps |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
View Code Duplication |
def pickle_to_s3(server_side_encryption=None, array_job=True): |
|
|
|
|
|
|
10
|
|
|
def real_decorator(f): |
|
11
|
|
|
@wraps(f) |
|
12
|
|
|
def wrapper(*args, **kwargs): |
|
13
|
|
|
s3 = boto3.client("s3") |
|
14
|
|
|
bucket = os.environ.get("CLOUDKNOT_JOBS_S3_BUCKET") |
|
15
|
|
|
|
|
16
|
|
|
if array_job: |
|
17
|
|
|
array_index = os.environ.get("AWS_BATCH_JOB_ARRAY_INDEX") |
|
18
|
|
|
else: |
|
19
|
|
|
array_index = '0' |
|
20
|
|
|
|
|
21
|
|
|
jobid = os.environ.get("AWS_BATCH_JOB_ID") |
|
22
|
|
|
|
|
23
|
|
|
if array_job: |
|
24
|
|
|
jobid = jobid.split(':')[0] |
|
25
|
|
|
|
|
26
|
|
|
key = '/'.join([ |
|
27
|
|
|
'cloudknot.jobs', |
|
28
|
|
|
os.environ.get("CLOUDKNOT_S3_JOBDEF_KEY"), |
|
29
|
|
|
jobid, |
|
30
|
|
|
array_index, |
|
31
|
|
|
'{0:03d}'.format(int(os.environ.get("AWS_BATCH_JOB_ATTEMPT"))), |
|
32
|
|
|
'output.pickle' |
|
33
|
|
|
]) |
|
34
|
|
|
|
|
35
|
|
|
result = f(*args, **kwargs) |
|
36
|
|
|
|
|
37
|
|
|
# Only pickle output and write to S3 if it is not None |
|
38
|
|
|
if result is not None: |
|
39
|
|
|
pickled_result = cloudpickle.dumps(result) |
|
40
|
|
|
if server_side_encryption is None: |
|
41
|
|
|
s3.put_object(Bucket=bucket, Body=pickled_result, Key=key) |
|
42
|
|
|
else: |
|
43
|
|
|
s3.put_object(Bucket=bucket, Body=pickled_result, Key=key, |
|
44
|
|
|
ServerSideEncryption=server_side_encryption) |
|
45
|
|
|
|
|
46
|
|
|
return wrapper |
|
47
|
|
|
return real_decorator |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
View Code Duplication |
def sensitivity_it(counter): |
|
|
|
|
|
|
51
|
|
|
|
|
52
|
|
|
import matplotlib as mpl |
|
53
|
|
|
mpl.use('Agg') |
|
54
|
|
|
import matplotlib.pyplot as plt |
|
55
|
|
|
import diff_classifier.aws as aws |
|
56
|
|
|
import diff_classifier.utils as ut |
|
57
|
|
|
import diff_classifier.msd as msd |
|
58
|
|
|
import diff_classifier.features as ft |
|
59
|
|
|
import diff_classifier.imagej as ij |
|
60
|
|
|
import diff_classifier.heatmaps as hm |
|
61
|
|
|
|
|
62
|
|
|
from scipy.spatial import Voronoi |
|
63
|
|
|
import scipy.stats as stats |
|
64
|
|
|
from shapely.geometry import Point |
|
65
|
|
|
from shapely.geometry.polygon import Polygon |
|
66
|
|
|
import matplotlib.cm as cm |
|
67
|
|
|
import os |
|
68
|
|
|
import os.path as op |
|
69
|
|
|
import numpy as np |
|
70
|
|
|
import numpy.ma as ma |
|
71
|
|
|
import pandas as pd |
|
72
|
|
|
import boto3 |
|
73
|
|
|
import itertools |
|
74
|
|
|
|
|
75
|
|
|
#Sweep parameters |
|
76
|
|
|
#---------------------------------- |
|
77
|
|
|
radius = [4.5, 6.0, 7.0] |
|
78
|
|
|
do_median_filtering = [True, False] |
|
79
|
|
|
quality = [1.5, 4.5, 8.5] |
|
80
|
|
|
linking_max_distance = [6.0, 10.0, 15.0] |
|
81
|
|
|
gap_closing_max_distance = [6.0, 10.0, 15.0] |
|
82
|
|
|
max_frame_gap = [1, 2, 5] |
|
83
|
|
|
track_displacement = [0.0, 10.0, 20.0] |
|
84
|
|
|
|
|
85
|
|
|
sweep = [radius, do_median_filtering, quality, linking_max_distance, gap_closing_max_distance, max_frame_gap, |
|
86
|
|
|
track_displacement] |
|
87
|
|
|
all_params = list(itertools.product(*sweep)) |
|
88
|
|
|
|
|
89
|
|
|
#Variable prep |
|
90
|
|
|
#---------------------------------- |
|
91
|
|
|
s3 = boto3.client('s3') |
|
92
|
|
|
|
|
93
|
|
|
folder = '01_18_Experiment' |
|
94
|
|
|
s_folder = '{}/sensitivity'.format(folder) |
|
95
|
|
|
local_folder = '.' |
|
96
|
|
|
prefix = "P1_S1_R_0001_2_2" |
|
97
|
|
|
name = "{}.tif".format(prefix) |
|
98
|
|
|
local_im = op.join(local_folder, name) |
|
99
|
|
|
aws.download_s3('{}/{}/{}.tif'.format(folder, prefix.split('_')[0], prefix), '{}.tif'.format(prefix)) |
|
100
|
|
|
|
|
101
|
|
|
outputs = np.zeros((len(all_params), len(all_params[0])+2)) |
|
102
|
|
|
|
|
103
|
|
|
#Tracking and calculations |
|
104
|
|
|
#------------------------------------ |
|
105
|
|
|
params = all_params[counter] |
|
106
|
|
|
outfile = 'Traj_{}_{}.csv'.format(name.split('.')[0], counter) |
|
107
|
|
|
msd_file = 'msd_{}_{}.csv'.format(name.split('.')[0], counter) |
|
108
|
|
|
geo_file = 'geomean_{}_{}.csv'.format(name.split('.')[0], counter) |
|
109
|
|
|
geoS_file = 'geoSEM_{}_{}.csv'.format(name.split('.')[0], counter) |
|
110
|
|
|
msd_image = 'msds_{}_{}.png'.format(name.split('.')[0], counter) |
|
111
|
|
|
iter_name = "{}_{}".format(prefix, counter) |
|
112
|
|
|
|
|
113
|
|
|
ij.track(local_im, outfile, template=None, fiji_bin=None, radius=params[0], threshold=0., |
|
114
|
|
|
do_median_filtering=params[1], quality=params[2], x=511, y=511, ylo=1, median_intensity=300.0, snr=0.0, |
|
115
|
|
|
linking_max_distance=params[3], gap_closing_max_distance=params[4], max_frame_gap=params[5], |
|
116
|
|
|
track_displacement=params[6]) |
|
117
|
|
|
|
|
118
|
|
|
traj = ut.csv_to_pd(outfile) |
|
119
|
|
|
msds = msd.all_msds2(traj, frames=651) |
|
120
|
|
|
msds.to_csv(msd_file) |
|
121
|
|
|
gmean1, gSEM1 = hm.plot_individual_msds(iter_name, alpha=0.05) |
|
122
|
|
|
np.savetxt(geo_file, gmean1, delimiter=",") |
|
123
|
|
|
np.savetxt(geoS_file, gSEM1, delimiter=",") |
|
124
|
|
|
|
|
125
|
|
|
aws.upload_s3(outfile, '{}/{}'.format(s_folder, outfile)) |
|
126
|
|
|
aws.upload_s3(msd_file, '{}/{}'.format(s_folder, msd_file)) |
|
127
|
|
|
aws.upload_s3(geo_file, '{}/{}'.format(s_folder, geo_file)) |
|
128
|
|
|
aws.upload_s3(geoS_file, '{}/{}'.format(s_folder, geoS_file)) |
|
129
|
|
|
aws.upload_s3(msd_image, '{}/{}'.format(s_folder, msd_image)) |
|
130
|
|
|
|
|
131
|
|
|
print('Successful parameter calculations for {}'.format(iter_name)) |
|
132
|
|
|
|
|
133
|
|
|
|
|
134
|
|
View Code Duplication |
if __name__ == "__main__": |
|
|
|
|
|
|
135
|
|
|
description = ('Download input from an S3 bucket and provide that input ' |
|
136
|
|
|
'to our function. On return put output in an S3 bucket.') |
|
137
|
|
|
|
|
138
|
|
|
parser = ArgumentParser(description=description) |
|
139
|
|
|
|
|
140
|
|
|
parser.add_argument( |
|
141
|
|
|
'bucket', metavar='bucket', type=str, |
|
142
|
|
|
help='The S3 bucket for pulling input and pushing output.' |
|
143
|
|
|
) |
|
144
|
|
|
|
|
145
|
|
|
parser.add_argument( |
|
146
|
|
|
'--starmap', action='store_true', |
|
147
|
|
|
help='Assume input has already been grouped into a single tuple.' |
|
148
|
|
|
) |
|
149
|
|
|
|
|
150
|
|
|
parser.add_argument( |
|
151
|
|
|
'--arrayjob', action='store_true', |
|
152
|
|
|
help='If True, this is an array job and it should reference the ' |
|
153
|
|
|
'AWS_BATCH_JOB_ARRAY_INDEX environment variable.' |
|
154
|
|
|
) |
|
155
|
|
|
|
|
156
|
|
|
parser.add_argument( |
|
157
|
|
|
'--sse', dest='sse', action='store', |
|
158
|
|
|
choices=['AES256', 'aws:kms'], default=None, |
|
159
|
|
|
help='Server side encryption algorithm used when storing objects ' |
|
160
|
|
|
'in S3.' |
|
161
|
|
|
) |
|
162
|
|
|
|
|
163
|
|
|
args = parser.parse_args() |
|
164
|
|
|
|
|
165
|
|
|
s3 = boto3.client('s3') |
|
166
|
|
|
bucket = args.bucket |
|
167
|
|
|
|
|
168
|
|
|
jobid = os.environ.get("AWS_BATCH_JOB_ID") |
|
169
|
|
|
|
|
170
|
|
|
if args.arrayjob: |
|
171
|
|
|
jobid = jobid.split(':')[0] |
|
172
|
|
|
|
|
173
|
|
|
key = '/'.join([ |
|
174
|
|
|
'cloudknot.jobs', |
|
175
|
|
|
os.environ.get("CLOUDKNOT_S3_JOBDEF_KEY"), |
|
176
|
|
|
jobid, |
|
177
|
|
|
'input.pickle' |
|
178
|
|
|
]) |
|
179
|
|
|
|
|
180
|
|
|
response = s3.get_object(Bucket=bucket, Key=key) |
|
181
|
|
|
input_ = pickle.loads(response.get('Body').read()) |
|
182
|
|
|
|
|
183
|
|
|
if args.arrayjob: |
|
184
|
|
|
array_index = int(os.environ.get("AWS_BATCH_JOB_ARRAY_INDEX")) |
|
185
|
|
|
input_ = input_[array_index] |
|
186
|
|
|
|
|
187
|
|
|
if args.starmap: |
|
188
|
|
|
pickle_to_s3(args.sse, args.arrayjob)(sensitivity_it)(*input_) |
|
189
|
|
|
else: |
|
190
|
|
|
pickle_to_s3(args.sse, args.arrayjob)(sensitivity_it)(input_) |
|
191
|
|
|
|
ccurtis7 /
diff_classifier
Loading history...