forked from daquang/PillowNet
-
Notifications
You must be signed in to change notification settings - Fork 0
/
predict.py
197 lines (171 loc) · 8.14 KB
/
predict.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
#!/usr/bin/env python
"""
Script for generating predictions from a trained model.
Use `predict.py -h` to see an auto-generated description of advanced options.
"""
import argparse
import numpy as np
from keras.models import load_model
from tqdm import tqdm, trange
import pybedtools as pbt
import pyBigWig as pbw
from genomeloader.wrapper import TwoBitWrapper, FastaWrapper, BedWrapper, BigWigWrapper
from genomeloader.generator import MultiBedGenerator
from pillownet.layer import ReverseComplement, Reverse
def get_args():
parser = argparse.ArgumentParser(description='Generating predictions.',
epilog='\n'.join(__doc__.strip().split('\n')[1:]).strip(),
formatter_class=argparse.RawTextHelpFormatter)
parser.add_argument('-w', '--weights', required=True,
help='Input model weights.', type=str)
parser.add_argument('-o', '--output', required=True,
help='Output bigWig file of predictions.', type=str)
parser.add_argument('-bl', '--blacklist', required=False,
default='resources/blacklist.bed.gz',
help='Blacklist BED file.', type=str)
parser.add_argument('-bw', '--bigwigs', type=str, required=False, nargs='*',
default=None,
help='Input bigwig files.')
parser.add_argument('-s', '--step',
help='Step size and window size to make predictions for.',
type=int, default=50)
parser.add_argument('-ch', '--channel',
help='If the model is multi-task, select which channel will be output (default: 0).',
type=int, default=0)
parser.add_argument('-t', '--threshold',
help='Remove all signal values below threshold (default: 1e-2).',
type=float, default=1e-2)
parser.add_argument('-at', '--autothreshold', action='store_true', default=False,
help='Automatically set threshold.')
parser.add_argument('-p', '--processes',
help='Number of parallel process workers (default: 3. If set to 0, then multiprocessing will '
'not be used).',
type=int, default=3)
group = parser.add_mutually_exclusive_group(required=False)
group.add_argument('-c', '--chroms', type=str, nargs='+',
default=['chr1', 'chr8', 'chr21'],
help='Chromosome(s) to make predictions for.')
group.add_argument('-wg', '--wholegenome', action='store_true', default=False,
help='Make predictions for the whole genome.')
group.add_argument('-ax', '--autox', action='store_true', default=False,
help='Predict on autosomes and X chromosome.')
group.add_argument('-a', '--auto', action='store_true', default=False,
help='Predict on autosomes.')
group = parser.add_mutually_exclusive_group(required=False)
group.add_argument('-gf', '--genomefasta', type=str,
help='Genome FASTA file.')
group.add_argument('-gt', '--genometwobit', type=str,
help='Genome twobit file.')
args = parser.parse_args()
return args
def main():
args = get_args()
weights_file = args.weights
output_file = args.output
step = args.step
channel = args.channel
workers = args.processes
threshold = args.threshold
autothreshold = args.autothreshold
bw = pbw.open(output_file, 'w')
if workers > 0:
use_multiprocessing = True
thread_safe = True
else:
workers = 0
use_multiprocessing = False
thread_safe = False
signals = []
if args.genometwobit is not None:
genome = TwoBitWrapper(args.genometwobit, thread_safe=thread_safe)
signals.append(genome)
elif args.genomefasta is not None:
genome = FastaWrapper(args.genomefasta, thread_safe=thread_safe)
signals.append(genome)
else:
genome = None
# Load bigwigs
bigwig_files = args.bigwigs
bigwigs = [] if bigwig_files is None else [BigWigWrapper(bigwig_file, thread_safe=thread_safe) for bigwig_file in
bigwig_files]
signals.extend(bigwigs)
chroms_size = signals[0].chroms_size()
# Load blacklist file
blacklist_file = args.blacklist
blacklist = None if blacklist_file is None else BedWrapper(blacklist_file)
if args.wholegenome:
chroms = signals[0].chroms()
elif args.autox:
chroms = ['chr1', 'chr10', 'chr11', 'chr12', 'chr13', 'chr14', 'chr15', 'chr16', 'chr17', 'chr18', 'chr19',
'chr2', 'chr20', 'chr21', 'chr22', 'chr3', 'chr4', 'chr5', 'chr6', 'chr7', 'chr8', 'chr9', 'chrX']
elif args.auto:
chroms = ['chr1', 'chr10', 'chr11', 'chr12', 'chr13', 'chr14', 'chr15', 'chr16', 'chr17', 'chr18', 'chr19',
'chr2', 'chr20', 'chr21', 'chr22', 'chr3', 'chr4', 'chr5', 'chr6', 'chr7', 'chr8', 'chr9']
else:
chroms = args.chroms
header = []
for chrom in chroms:
chrom_size = chroms_size[chrom]
header.append((chrom, chrom_size))
bw.addHeader(header)
model = load_model(weights_file, custom_objects={'ReverseComplement': ReverseComplement,
'Reverse': Reverse}, compile=False)
if autothreshold:
input_shape = model.input_shape
if type(input_shape) is list:
input_shape = [np.array(i) for i in input_shape]
for i in input_shape:
i[0] = 1
input_zeros = [np.zeros(i) for i in input_shape]
else:
input_shape = np.array(input_shape)
input_shape[0] = 1
input_zeros = np.zeros(input_shape)
output_zeros = model.predict(input_zeros)
threshold = output_zeros.max() * 1.01
print('The new threshold is: %f' % threshold)
return_sequences = len(model.output_shape) == 3
multi_task = model.output_shape[-1] > 1
dna_input_shape = model.input_shape[0] if isinstance(model.input_shape, list) else model.input_shape
seq_len = dna_input_shape[1]
output_seq_len = None
if return_sequences:
step = model.output_shape[1]
output_seq_len = model.output_shape[1]
pbar = tqdm(chroms)
for chrom in pbar:
pbar.set_description('Processing %s' % chrom)
chrom_size = chroms_size[chrom]
chrom_windows_bt = pbt.BedTool().window_maker(genome={chrom: (0, chrom_size)}, w=step, s=step)
chrom_windows = BedWrapper(chrom_windows_bt.fn, sort_bed=False)
generator = MultiBedGenerator(beds=[chrom_windows], signals=signals, seq_len=seq_len,
output_seq_len=output_seq_len, negatives_ratio=0, jitter_mode=None,
shuffle=False, return_sequences=return_sequences, return_output=False,
left_justify=True)
chrom_start = 0
for i in trange(len(generator)):
batch = generator[i]
#chrom_end = chrom_start + step * len(batch)
predictions_batch = model.predict(batch)
if multi_task:
predictions_batch = predictions_batch[:, :, channel]
values = predictions_batch.ravel()
chrom_end = chrom_start + len(values)
starts = np.arange(chrom_start, chrom_end)
if blacklist is not None:
values_blacklist = blacklist[chrom, chrom_start:chrom_end].ravel()
values[values_blacklist] = 0
if chrom_end > chrom_size:
crop_size = chrom_size - chrom_end
values = values[:crop_size]
starts = starts[:crop_size]
chrom_start = chrom_end
above_threshold = values >= threshold
values = values[above_threshold]
if len(values) == 0:
continue
starts = starts[above_threshold]
bw.addEntries(chroms=chrom, starts=starts, span=1, values=values)
bw.close()
if __name__ == '__main__':
main()