-
Notifications
You must be signed in to change notification settings - Fork 37
/
stitch_utils.py
189 lines (167 loc) · 6.57 KB
/
stitch_utils.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
# Copyright (c) 2021, Google Inc.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without modification,
# are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of Google Inc. nor the names of its contributors
# may be used to endorse or promote products derived from this software without
# specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR
# ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
# (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
# LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
# ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Methods for DeepConsensus stitch-predictions step."""
import dataclasses
from typing import Iterable, Optional, Tuple
from absl import logging
import numpy as np
from deepconsensus.utils import dc_constants
from deepconsensus.utils import utils
@dataclasses.dataclass
class DCModelOutput:
molecule_name: str
window_pos: int
ec: float
np_num_passes: int
rq: float
rg: str
sequence: Optional[str] = None
quality_string: Optional[str] = None
def get_full_sequence(
deepconsensus_outputs: Iterable[DCModelOutput],
max_length: int,
fill_n: bool = False,
):
"""Stitch together windows of predictions into a full sequence."""
# Build up the full sequence from the sorted windows.
full_sequence_parts = []
quality_string_parts = []
start = 0
# DeepConsensus outputs are expected to be sorted.
for dc_output in deepconsensus_outputs:
# This while loop is used to handle missing windows
while dc_output.window_pos > start:
if not fill_n:
return None, ''
else:
# Add N-base filler for sequences that were unable to be inferred.
full_sequence_parts.append('N' * max_length)
empty_quality_scores = np.array([dc_constants.EMPTY_QUAL] * max_length)
empty_quality_string = utils.quality_scores_to_string(
empty_quality_scores
)
quality_string_parts.append(empty_quality_string)
start += max_length
full_sequence_parts.append(dc_output.sequence)
quality_string_parts.append(dc_output.quality_string)
start += max_length
full_sequence = ''.join(full_sequence_parts)
full_quality_string = ''.join(quality_string_parts)
return full_sequence, full_quality_string
def remove_gaps(sequence: str, quality_string: str) -> Tuple[str, str]:
"""Removes gaps and corresponding quality score from outputs."""
# Remove gaps from the final sequence.
final_sequence = ''
final_quality_string = ''
bases_to_remove = set([dc_constants.GAP])
# Only keep bases and quality scores for non gap positions.
for base, quality in zip(sequence, quality_string):
if base not in bases_to_remove:
final_sequence += base
final_quality_string += quality
assert len(final_sequence) == len(final_quality_string)
assert dc_constants.GAP not in final_sequence
return final_sequence, final_quality_string
def is_quality_above_threshold(quality_string, min_quality):
quality_score_array = utils.quality_string_to_array(quality_string)
# Round the phred score to ensure expected behavior. Without rounding, a
# read with all base qualities equal to 10 will have an average phred of
# 9.99999 due to python floating point precision. Such as read would get
# filtered out if min_quality is 10.
rounded_avg_phred = round(utils.avg_phred(quality_score_array), 5)
logging.vlog(3, 'Quality is %d', rounded_avg_phred)
return rounded_avg_phred >= min_quality
def format_as_fastq(
molecule_name: str, sequence: str, quality_string: str
) -> str:
formatted_for_fastq = f'@{molecule_name}\n'
formatted_for_fastq += f'{sequence}\n'
formatted_for_fastq += '+\n'
formatted_for_fastq += f'{quality_string}\n'
return formatted_for_fastq
@dataclasses.dataclass
class OutcomeCounter:
empty_sequence: int = 0
only_gaps: int = 0
failed_quality_filter: int = 0
failed_length_filter: int = 0
success: int = 0
def stitch_to_fastq(
molecule_name: str,
predictions: Iterable[DCModelOutput],
max_length: int,
min_quality: int,
min_length: int,
outcome_counter: OutcomeCounter,
) -> Optional[str]:
"""Stitch windows of predictions together, filter, and make FASTQ string."""
full_sequence, full_quality_string = get_full_sequence(
deepconsensus_outputs=predictions, max_length=max_length
)
# Filter out the read if it is empty after stitching.
if not full_sequence:
outcome_counter.empty_sequence += 1
logging.vlog(
1, 'Filtered out read that was empty after stitching: %s', molecule_name
)
return None
final_sequence, final_quality_string = remove_gaps(
sequence=full_sequence, quality_string=full_quality_string
)
# Filter out the read if it contains only gaps and no bases.
if not final_sequence:
outcome_counter.only_gaps += 1
logging.vlog(
1,
'Filtered out read that contained only gaps and no bases: %s',
molecule_name,
)
return None
# Filter out the read if its quality scores are too low.
if not is_quality_above_threshold(
quality_string=final_quality_string, min_quality=min_quality
):
outcome_counter.failed_quality_filter += 1
logging.vlog(
1, 'Filtered out read below quality threshold: %s', molecule_name
)
return None
# Filter out the read if it is too short.
if len(final_sequence) < min_length:
outcome_counter.failed_length_filter += 1
logging.vlog(
1, 'Filtered out read below length threshold: %s', molecule_name
)
return None
fastq = format_as_fastq(
molecule_name=molecule_name,
sequence=final_sequence,
quality_string=final_quality_string,
)
outcome_counter.success += 1
return fastq