-
Notifications
You must be signed in to change notification settings - Fork 1
/
pathogen_properties.py
executable file
·543 lines (440 loc) · 17.8 KB
/
pathogen_properties.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
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
import abc
import calendar
import dataclasses
import datetime
import itertools
import os.path
import re
from collections.abc import Iterable
from dataclasses import InitVar, dataclass, field
from enum import Enum
from typing import NewType, Optional
import numpy as np
# Enums, short for enumerations, are a data type in Python used to represent a set of named values,
# which are typically used to define a set of related constants with unique names.
# TODO: Potentially create enum for country, state, etc. to ensure consistency in naming
class NAType(Enum):
DNA = "DNA"
RNA = "RNA"
class SelectionRound(Enum):
ROUND_1 = "Round 1" # Public health only
ROUND_2 = "Round 2" # Considering sequencing results
class Enveloped(Enum):
ENVELOPED = "enveloped"
NON_ENVELOPED = "non_enveloped"
class VariableType(Enum):
MEASUREMENT = "measurement"
EXTERNAL_ESTIMATE = "external_estimate"
NAO_ESTIMATE = "nao_estimate"
class Active(Enum):
ACTIVE = "Active"
LATENT = "Latent"
TaxID = NewType("TaxID", int)
@dataclass(kw_only=True, eq=True, frozen=True)
class PathogenChars:
na_type: NAType
enveloped: Enveloped
selection: SelectionRound
# Set exactly one of taxid or taxids; read taxids.
#
# Normally you only should set taxid. Set taxids in cases like the flu
# where surveillance generally conflates Flu A and Flu B but they don't
# form a clade.
taxid: InitVar[Optional[TaxID]] = None
taxids: Optional[frozenset[TaxID]] = None
# If we produce any estimates more specific than the overall taxid,
# subtaxids will contain all the secondary taxonomic ids we can generate.
subtaxids: frozenset[TaxID] = frozenset()
names_by_taxid: Optional[dict[TaxID, str]] = None
def __post_init__(self, taxid: Optional[TaxID]):
assert bool(taxid) ^ bool(self.taxids) # Exactly one should be set.
if taxid:
# A python wart is that frozen dataclasses don't have an exception
# for __post_init__, and it thinks assignments here are mutation
# instead of initialization. That's why we're assigning with
# __setattr__. See
# https://stackoverflow.com/questions/53756788/how-to-set-the-value-of-dataclass-field-in-post-init-when-frozen-true/54119384#54119384
object.__setattr__(self, "taxids", frozenset([taxid]))
def days_in_month(year: int, month: int) -> int:
_, last_day = calendar.monthrange(year, month)
return last_day
@dataclass(kw_only=True, eq=True, frozen=True)
class Variable:
"""An external piece of data"""
source: Optional[str] = None
country: Optional[str] = None
state: Optional[str] = None
county: Optional[str] = None
number_of_participants: Optional[int] = None
confidence_interval: Optional[tuple[float, float]] = None
coverage_probability: Optional[float] = None
methods: Optional[str] = None
# Either supply date, or start_date and end_date.
# Dates can be any of: YYYY, YYYY-MM, or YYYY-MM-DD.
date: InitVar[Optional[str]] = None
start_date: InitVar[Optional[str]] = None
end_date: InitVar[Optional[str]] = None
# In cases where an estimate is derived from multiple input variables with
# different dates, set date_source to the Variable that represents the date
# range this estimate is intended for. For example, imagine we have:
#
# population = Population(date="2020-04-01", ...)
# prevalence = AbsolutePrevalence(date="2020-08-01", ...)
# return prevalence.to_rate(population)
#
# There's no way for the consumer to know what date this estimate is for.
# So instead we do:
#
# return prevalence.to_rate(population, date_source=prevalence)
#
# parsed_start / parsed_end are set from date_source if supplied, otherwise
# from start_date / end_date.
date_source: InitVar[Optional["Variable"]] = None
# Same deal as date_source: set this if otherwise it wouldn't be clear
# what location an estimate would be for.
location_source: InitVar[Optional["Variable"]] = None
# Read these via get_dates(), which asserts that they're set.
parsed_start: Optional[datetime.date] = None
parsed_end: Optional[datetime.date] = None
taxid: Optional[TaxID] = None
inputs: InitVar[Optional[Iterable["Variable"]]] = None
all_inputs: set["Variable"] = field(default_factory=set)
is_pseudocount: Optional[bool] = None
def __post_init__(
self,
date: Optional[str],
start_date: Optional[str],
end_date: Optional[str],
date_source: Optional["Variable"],
location_source: Optional["Variable"],
inputs: Optional[Iterable["Variable"]],
):
# See comment above about __post_init__ for why we're using
# __setattr__.
if date and (start_date or end_date):
raise Exception("If you have start/end don't set date.")
if self.parsed_start and (date or start_date):
raise Exception("Don't set both parsed_start and provide a date")
if self.parsed_end and (date or end_date):
raise Exception("Don't set both parsed_start and provide a date")
if (start_date and not end_date) or (end_date and not start_date):
raise Exception("Start and end must go together.")
if date:
start_date = end_date = date
parsed_start = self.parsed_start
if start_date:
parsed_start = self._parse_date(start_date, "start")
parsed_end = self.parsed_end
if end_date:
parsed_end = self._parse_date(end_date, "end")
if date_source:
assert date_source.parsed_start
assert date_source.parsed_end
parsed_start = date_source.parsed_start
parsed_end = date_source.parsed_end
if parsed_start and parsed_end and parsed_start > parsed_end:
raise Exception("Start date can't be after end date")
object.__setattr__(self, "parsed_start", parsed_start)
object.__setattr__(self, "parsed_end", parsed_end)
if location_source:
object.__setattr__(self, "country", location_source.country)
object.__setattr__(self, "state", location_source.state)
object.__setattr__(self, "county", location_source.county)
elif inputs:
# If they didn't give us location information but there's location
# information in our inputs, check that for consistency. If it's
# consistent use it, otherwise raise an error.
countries = set(i.country for i in inputs if i.country)
states = set(i.state for i in inputs if i.state)
counties = set(i.county for i in inputs if i.county)
country = self.country
state = self.state
county = self.county
if not country:
(country,) = countries
if states and not state:
(state,) = states
if counties and not county:
(county,) = counties
object.__setattr__(self, "country", country)
object.__setattr__(self, "state", state)
object.__setattr__(self, "county", county)
all_inputs = set(self.all_inputs or inputs or [])
if date_source:
all_inputs.add(date_source)
if location_source:
all_inputs.add(location_source)
for variable in list(all_inputs):
all_inputs |= variable.all_inputs
object.__setattr__(self, "all_inputs", frozenset(all_inputs))
def _parse_date(self, date: str, start_or_end: str) -> datetime.date:
y, m, d = None, None, None
if y_match := re.findall("^(\d\d\d\d)$", date):
(y,) = y_match
elif ym_match := re.findall("^(\d\d\d\d)-(\d\d)$", date):
((y, m),) = ym_match
elif ymd_match := re.findall("^(\d\d\d\d)-(\d\d)-(\d\d)$", date):
((y, m, d),) = ymd_match
else:
raise Exception("Unknown date format %s" % date)
y = int(y)
if m:
m = int(m)
else:
m = {"start": 1, "end": 12}[start_or_end]
if d:
d = int(d)
else:
if start_or_end == "start":
d = 1
else:
d = days_in_month(int(y), int(m))
return datetime.date(y, m, d)
def get_dates(self) -> tuple[datetime.date, datetime.date]:
assert self.parsed_start
assert self.parsed_end
return self.parsed_start, self.parsed_end
def get_date(self) -> datetime.date:
# Only call this on variables that you know represent a single-day
# estimate.
start, end = self.get_dates()
assert start == end
return start
def get_location(
self,
) -> tuple[Optional[str], Optional[str], Optional[str]]:
return self.country, self.state, self.county
def summarize_location(self) -> str:
country, state, county = self.get_location()
return ", ".join(x for x in [county, state, country] if x)
@staticmethod
def _weightedAverageByPopulation(
*pairs: tuple[float, "Population"]
) -> float:
return float(
np.average(
[val for (val, population) in pairs],
weights=[population.people for (val, population) in pairs],
)
)
@dataclass(kw_only=True, eq=True, frozen=True)
class Taggable(Variable):
# In cases where the location and date isn't enough to identify the
# population, you can set a more specific tag to reduce errors. For
# example, tag="18-49yo".
tag: Optional[str] = None
def assert_comparable(self, other: "Taggable"):
v1 = self
v2 = other
assert v1.country == v2.country
assert v1.state == v2.state
assert v1.county == v2.county
# Normally everything has to match, but it's ok if one of them
# has a more specific date as long as it's within a year; populations
# don't change quickly.
v1_start, v1_end = v1.get_dates()
v2_start, v2_end = v2.get_dates()
assert v1_start.year == v2_start.year
assert v1_end.year == v2_end.year
assert v1.tag == v2.tag
@dataclass(kw_only=True, eq=True, frozen=True)
class Scalar(Variable):
scalar: float
@staticmethod
def average(in1: "Scalar", in2: "Scalar"):
return Scalar(scalar=(in1.scalar + in2.scalar) / 2, inputs=[in1, in2])
class Predictor(abc.ABC, Variable):
@abc.abstractmethod
def get_data(self) -> float: ...
@dataclass(kw_only=True, eq=True, frozen=True)
class Population(Taggable):
"""A number of people"""
people: float
def __mul__(self, scalar: Scalar) -> "Population":
return Population(
people=self.people * scalar.scalar,
inputs=[self, scalar],
date_source=self,
location_source=self,
)
def __sub__(self: "Population", other: "Population") -> "Population":
return Population(
people=self.people - other.people,
inputs=[self, other],
date_source=self,
location_source=self,
)
@dataclass(kw_only=True, eq=True, frozen=True)
class Prevalence(Predictor):
"""What fraction of people have this pathogen at some moment"""
active: Active
infections_per_100k: float
def get_data(self) -> float:
return self.infections_per_100k
def __mul__(self, scalar: Scalar) -> "Prevalence":
return Prevalence(
infections_per_100k=self.infections_per_100k * scalar.scalar,
inputs=[self, scalar],
active=self.active,
date_source=self,
location_source=self,
)
def __truediv__(self, scalar: Scalar) -> "Prevalence":
return Prevalence(
infections_per_100k=self.infections_per_100k / scalar.scalar,
inputs=[self, scalar],
active=self.active,
date_source=self,
location_source=self,
)
def __add__(self: "Prevalence", other: "Prevalence") -> "Prevalence":
assert self.active == other.active
assert self.parsed_start == other.parsed_start
assert self.parsed_end == other.parsed_end
return Prevalence(
infections_per_100k=self.infections_per_100k
+ other.infections_per_100k,
inputs=[self, other],
active=self.active,
date_source=self,
location_source=self,
)
@staticmethod
def weightedAverageByPopulation(
*pairs: tuple["Prevalence", "Population"]
) -> "Prevalence":
return dataclasses.replace(
pairs[0][0],
infections_per_100k=Variable._weightedAverageByPopulation(
*[
(prevalence.infections_per_100k, population)
for (prevalence, population) in pairs
]
),
location_source=pairs[0][1],
date_source=pairs[0][1],
inputs=itertools.chain.from_iterable(pairs),
)
def __sub__(self: "Prevalence", other: "Prevalence") -> "Prevalence":
assert self.active == other.active
assert self.parsed_start == other.parsed_start
assert self.parsed_end == other.parsed_end
return Prevalence(
infections_per_100k=self.infections_per_100k
- other.infections_per_100k,
inputs=[self, other],
active=self.active,
date_source=self,
location_source=self,
)
@dataclass(kw_only=True, eq=True, frozen=True)
class PrevalenceAbsolute(Taggable):
"""How many people had this pathogen at some moment"""
infections: float
active: Active
def __mul__(self, scalar: Scalar) -> "PrevalenceAbsolute":
return PrevalenceAbsolute(
infections=self.infections * scalar.scalar,
inputs=[self, scalar],
date_source=self,
location_source=self,
active=self.active,
)
def to_rate(self, population: Population) -> Prevalence:
self.assert_comparable(population)
return Prevalence(
infections_per_100k=self.infections * 100000 / population.people,
inputs=[self, population],
active=self.active,
date_source=self,
location_source=self,
)
@dataclass(kw_only=True, eq=True, frozen=True)
class Number(Variable):
"""Generic number. Use this for weird one-off things
If some concept is being used by more than two estimates it should get a
more specific Variable subclass."""
number: float
def __truediv__(self, other: "Number") -> Scalar:
return Scalar(
scalar=self.number / other.number,
inputs=[self, other],
date_source=self,
location_source=self,
)
@dataclass(kw_only=True, eq=True, frozen=True)
class IncidenceRate(Predictor):
"""What fraction of people get this pathogen annually"""
annual_infections_per_100k: float
def get_data(self) -> float:
return self.get_weekly_infections_per_100k()
def get_weekly_infections_per_100k(self) -> float:
return self.annual_infections_per_100k / 52
def __mul__(self, scalar: Scalar) -> "IncidenceRate":
return IncidenceRate(
annual_infections_per_100k=self.annual_infections_per_100k
* scalar.scalar,
inputs=[self, scalar],
date_source=self,
location_source=self,
)
@staticmethod
def weightedAverageByPopulation(
*pairs: tuple["IncidenceRate", "Population"]
) -> "IncidenceRate":
return dataclasses.replace(
pairs[0][0],
annual_infections_per_100k=Variable._weightedAverageByPopulation(
*[
(incidence.annual_infections_per_100k, population)
for (incidence, population) in pairs
]
),
location_source=pairs[0][1],
date_source=pairs[0][1],
inputs=itertools.chain.from_iterable(pairs),
)
@dataclass(kw_only=True, eq=True, frozen=True)
class IncidenceAbsolute(Taggable):
"""How many people get this pathogen annually"""
annual_infections: float
def to_rate(self, population: Population) -> IncidenceRate:
self.assert_comparable(population)
return IncidenceRate(
annual_infections_per_100k=self.annual_infections
* 100000
/ population.people,
inputs=[self, population],
date_source=self,
location_source=self,
)
def __truediv__(self, other: "IncidenceAbsolute"):
assert self.tag == other.tag
return Scalar(
scalar=self.annual_infections / other.annual_infections,
inputs=[self, other],
)
def prevalence_data_filename(filename):
return os.path.join(os.path.dirname(__file__), "prevalence-data", filename)
def by_taxids(
pathogen_chars: PathogenChars, predictors: list[Predictor]
) -> dict[frozenset[TaxID], list[Predictor]]:
out: dict[frozenset[TaxID], list[Predictor]] = {}
for predictor in predictors:
taxids = pathogen_chars.taxids
if predictor.taxid:
taxids = frozenset([predictor.taxid])
assert taxids
if taxids not in out:
out[taxids] = []
out[taxids].append(predictor)
return out
# We don't want to predict zero of any pathogen, both because they almost never
# go truely to zero and because modeling will be taking logs of our output.
# Instead, consider our limit of detection to be a single event (an outbreak or
# reported postive test), and when we have no recorded events figure there were
# actually 0.1.
#
# It might be better to handle this in the modeling step, but by the time we
# get to that point the granularity of the input data has been discarded.
QUANTITY_WHEN_NONE_OBSERVED = 0.1