forked from GuillaumeDerval/newcovid_belgium
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathanalysis_VOCs_GISAID records_Thailand.R
574 lines (499 loc) · 35.2 KB
/
analysis_VOCs_GISAID records_Thailand.R
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
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
# ANALYSIS OF GROWTH ADVANTAGE OF DIFFERENT VOCs IN THAILAND (GISAID RECORDS)
# T. Wenseleers
# last update 7 OCTOBER 2021
library(nnet)
# devtools::install_github("melff/mclogit",subdir="pkg") # install latest development version of mclogit, to add emmeans support
library(mclogit)
# remotes::install_github("rvlenth/emmeans", dependencies = TRUE, force = TRUE)
library(emmeans)
library(readr)
library(ggplot2)
library(ggthemes)
library(scales)
library(stringr)
library(lubridate)
today = as.Date(Sys.time()) # we use the file date version as our definition of "today"
today = as.Date("2021-10-08")
today_num = as.numeric(today)
plotdir = "Thailand_GISAID"
suppressWarnings(dir.create(paste0(".//plots//",plotdir)))
# import GISAID records for Thailand
d1 = read_tsv(file(".//data//GISAID//Thailand//gisaid_hcov-19_2021_10_08_08_subm_2020_oct_2021.tsv"), col_types = cols(.default = "c"))
GISAID = as.data.frame(rbind(d1))
colnames(GISAID) = c("Virus name","Accession ID","date","Location","host",
"Additional location information","Sampling strategy","Gender",
"Patient age","Patient status","Last vaccinated","Passage","Specimen",
"Additional host information","pango_lineage","Clade","AA Substitutions")
date_isvalid = sapply(GISAID$date, function (s) str_count(s, pattern = "-")==2)
GISAID = GISAID[date_isvalid,]
GISAID$date = as.Date(GISAID$date)
GISAID = GISAID[!is.na(GISAID$date),]
GISAID = GISAID[GISAID$host=="Human",]
GISAID = GISAID[GISAID$date>=as.Date("2021-01-01"),]
range(GISAID$date) # "2021-01-01" "2021-09-18"
GISAID$Week = lubridate::week(GISAID$date)
GISAID$Year = lubridate::year(GISAID$date)
GISAID$Year_Week = interaction(GISAID$Year,GISAID$Week)
GISAID$floor_date = as.Date(as.character(cut(GISAID$date, "week")))+3.5 # week midpoint date
GISAID$DATE_NUM = as.numeric(GISAID$date)
GISAID = GISAID[GISAID$pango_lineage!="None",]
GISAID$pango_lineage[grepl("B.1.177",GISAID$pango_lineage,fixed=T)] = "B.1.177+"
GISAID$pango_lineage[grepl("B.1.621",GISAID$pango_lineage,fixed=T)] = "B.1.621+"
GISAID$pango_lineage[grepl("B.1.36\\>",GISAID$pango_lineage)] = "B.1.36+"
# GISAID$pango_lineage[grepl("B.1.617.2|AY",GISAID$pango_lineage)] = "Delta (B.1.617.2 & AY.X)"
# sel_target_VOC = "Delta (B.1.617.2 & AY.X)"
sel_target_VOC = "B.1.617.2"
GISAID$LINEAGE = GISAID$pango_lineage
nrow(GISAID) # 2736
# ANALYSIS OF VOCs IN THAILAND ####
# sel_countries = "Thailand"
# GISAID[GISAID$country %in% sel_countries,]
GISAID_sel = GISAID
sum(GISAID_sel$LINEAGE==sel_target_VOC) # 177
table(GISAID_sel$LINEAGE)
main_lineages = names(table(GISAID_sel$LINEAGE))[100*table(GISAID_sel$LINEAGE)/sum(table(GISAID_sel$LINEAGE)) > 1]
main_lineages
# "AY.30" "B.1.1.7" "B.1.351" "B.1.36+" "B.1.617.2"
VOCs = c("B.1.617.1","B.1.617.2","B.1.617+","B.1.618","B.1.1.7","B.1.351","P.1","B.1.1.318","B.1.1.207","B.1.429",
"B.1.525","B.1.526","B.1.1.519","B.1.1.318","B.1.621+",sel_target_VOC)
main_lineages = union(main_lineages, VOCs)
GISAID_sel$LINEAGE[!(GISAID_sel$LINEAGE %in% main_lineages)] = "other" # minority lineages & non-VOCs
remove = names(table(GISAID_sel$LINEAGE))[table(GISAID_sel$LINEAGE)/sum(table(GISAID_sel$LINEAGE)) < 0.01]
remove = remove[!(remove %in% c("B.1.351","B.1.1.7","P.1","B.1.617.2","B.1.617.1","B.1.1.519","B.1.621+",sel_target_VOC,"B.1.351","P.1","AY.23","AY.30","AY.4"))]
GISAID_sel$LINEAGE[(GISAID_sel$LINEAGE %in% remove)] = "other" # minority VOCs
table(GISAID_sel$LINEAGE)
GISAID_sel$LINEAGE = factor(GISAID_sel$LINEAGE)
GISAID_sel$LINEAGE = relevel(GISAID_sel$LINEAGE, ref="B.1.1.7") # we code UK strain as the reference level
levels(GISAID_sel$LINEAGE)
# "B.1.1.7" "AY.30" "B.1.351" "B.1.36+" "B.1.617.1" "B.1.617.2" "other"
levels_LINEAGE = c("B.1.1.7","B.1.351","B.1.36+","B.1.617.1",
sel_target_VOC,"AY.30","other")
GISAID_sel$LINEAGE = factor(GISAID_sel$LINEAGE, levels=levels_LINEAGE)
# aggregated data to make Muller plots of raw data
# aggregate by day to identify days on which INSA performed (days with a lot of sequences)
# we subset the data to just those days to avoid sampling biases (delta infection clusters etc)
data_agbyday = as.data.frame(table(GISAID_sel$date, GISAID_sel$LINEAGE))
colnames(data_agbyday) = c("date", "LINEAGE", "count")
data_agbyday_sum = aggregate(count ~ date, data=data_agbyday, sum)
data_agbyday$total = data_agbyday_sum$count[match(data_agbyday$date, data_agbyday_sum$date)]
sum(data_agbyday[data_agbyday$LINEAGE=="P.1","total"]) == nrow(GISAID_sel) # correct
data_agbyday$date = as.Date(as.character(data_agbyday$date))
data_agbyday$LINEAGE = factor(data_agbyday$LINEAGE, levels=levels_LINEAGE)
data_agbyday$date_num = as.numeric(data_agbyday$date)
data_agbyday$prop = data_agbyday$count/data_agbyday$total
data_agbyday$floor_date = NULL
# qplot(data=data_agbyday, x=date, y=total, colour=total>20, fill=total>20, geom="col")
GISAID_sel$total_sequenced_on_that_day = data_agbyday$total[match(GISAID_sel$date, data_agbyday$date)]
# GISAID_sel = GISAID_sel[GISAID_sel$total_sequenced_on_that_day>20,] # dates on which was performed
# nrow(GISAID_sel) # 5710
# aggregated by week for selected variant lineages
data_agbyweek = as.data.frame(table(GISAID_sel$floor_date, GISAID_sel$LINEAGE))
colnames(data_agbyweek) = c("floor_date", "LINEAGE", "count")
data_agbyweek_sum = aggregate(count ~ floor_date, data=data_agbyweek, sum)
data_agbyweek$total = data_agbyweek_sum$count[match(data_agbyweek$floor_date, data_agbyweek_sum$floor_date)]
sum(data_agbyweek[data_agbyweek$LINEAGE=="P.1","total"]) == nrow(GISAID_sel) # correct
data_agbyweek$collection_date = as.Date(as.character(data_agbyweek$floor_date))
data_agbyweek$LINEAGE = factor(data_agbyweek$LINEAGE, levels=levels_LINEAGE)
data_agbyweek$collection_date_num = as.numeric(data_agbyweek$collection_date)
data_agbyweek$prop = data_agbyweek$count/data_agbyweek$total
data_agbyweek$floor_date = NULL
# MULLER PLOT (RAW DATA)
n2 = length(levels(GISAID_sel$LINEAGE))
lineage_cols2 = hcl(h = seq(15, 320, length = n2), l = 65, c = 200)
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="B.1.1.7")] = "#0085FF"
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="B.1.351")] = "#9A9D00"
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="B.1.177+")] = "grey55"
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="P.1")] = "cyan3"
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="B.1.617.1")] = muted("magenta")
lineage_cols2[which(levels(GISAID_sel$LINEAGE)==sel_target_VOC)] = "magenta"
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="AY.4")] = muted("red",c=150,l=60)
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="AY.23")] = muted("red",c=150,l=40)
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="AY.30")] = "red"
lineage_cols2[which(levels(GISAID_sel$LINEAGE)=="other")] = "grey75"
muller_thailand_raw2 = ggplot(data=data_agbyweek, aes(x=collection_date, y=count, group=LINEAGE)) +
# facet_wrap(~ STATE, ncol=1) +
# geom_col(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE1), width=1, position="fill") +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE, group=LINEAGE), position="fill") +
scale_fill_manual("", values=lineage_cols2) +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01","2021-10-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01","2021-10-01"))),1,1),
limits=as.Date(c("2021-01-01",NA)),
expand=c(0,0)) +
# guides(color = guide_legend(reverse=F, nrow=2, byrow=T), fill = guide_legend(reverse=F, nrow=2, byrow=T)) +
theme_hc() +
# labs(title = "MAIN SARS-CoV2 VARIANT LINEAGES IN THE UK") +
ylab("Share") +
theme(legend.position="right",
axis.title.x=element_blank()) +
labs(title = "SPREAD OF SARS-CoV2 VARIANTS OF CONCERN IN THAILAND\n(GISAID data)")
# +
# coord_cartesian(xlim=c(1,max(GISAID_sel$Week)))
muller_thailand_raw2
ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_muller plots_raw data.png"), width=8, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_muller plots_raw data.pdf"), width=8, height=6)
# multinomial fits
data_agbyweek$LINEAGE = relevel(data_agbyweek$LINEAGE, ref=sel_target_VOC)
data_agbyweek$DATE_NUM = as.numeric(data_agbyweek$collection_date)
library(nnet)
library(splines)
set.seed(1)
fit1_thailand_multi = nnet::multinom(LINEAGE ~ scale(DATE_NUM), weights=count, data=data_agbyweek, maxit=1000)
fit2_thailand_multi = nnet::multinom(LINEAGE ~ ns(DATE_NUM, df=2), weights=count, data=data_agbyweek, maxit=1000)
BIC(fit1_thailand_multi, fit2_thailand_multi)
# df BIC
# fit1_thailand_multi 12 4929.478
# fit2_thailand_multi 18 4667.166
# growth rate advantage compared to Delta (difference in growth rate per day)
emtrthailand = emtrends(fit2_thailand_multi, trt.vs.ctrl ~ LINEAGE,
var="DATE_NUM", mode="latent",
at=list(DATE_NUM=max(GISAID_sel$DATE_NUM)),
adjust="none", df=NA)
delta_r_thailand = data.frame(confint(emtrthailand,
adjust="none", df=NA)$contrasts,
p.value=as.data.frame(emtrthailand$contrasts,
adjust="none", df=NA)$p.value)
delta_r_thailand
# contrast estimate SE df asymp.LCL asymp.UCL p.value
# 1 B.1.1.7 - B.1.617.2 -0.07145868 0.004507305 NA -0.08029284 -0.062624528 1.319720e-56
# 2 B.1.351 - B.1.617.2 -0.04000469 0.006202810 NA -0.05216198 -0.027847410 1.122591e-10
# 3 (B.1.36+) - B.1.617.2 -0.16301173 0.032470821 NA -0.22665337 -0.099370095 5.160349e-07
# 4 B.1.617.1 - B.1.617.2 -0.60088443 0.697180867 NA -1.96733382 0.765564956 3.887550e-01
# 5 AY.30 - B.1.617.2 -0.06311766 0.004868610 NA -0.07265996 -0.053575361 1.952450e-38
# 6 other - B.1.617.2 -0.01858785 0.005195245 NA -0.02877035 -0.008405359 3.464208e-04
# AY.30 35% less infectious than B1.1.617.2 with GT of 4.7 days
exp(0.06311766 *4.7) # = 1.345
# fitted prop of different LINEAGES in the thailand today
multinom_preds_today_avg = data.frame(emmeans(fit2_thailand_multi, ~ LINEAGE|1,
at=list(DATE_NUM=today_num),
mode="prob", df=NA))
multinom_preds_today_avg
# LINEAGE prob SE df asymp.LCL asymp.UCL
# 1 B.1.617.2 8.140261e-01 3.885975e-02 NA 7.378623e-01 8.901898e-01
# 2 B.1.1.7 1.311293e-02 3.903230e-03 NA 5.462736e-03 2.076312e-02
# 3 B.1.351 1.368937e-02 6.617819e-03 NA 7.186811e-04 2.666005e-02
# 4 B.1.36+ 2.956255e-10 1.239939e-09 NA -2.134610e-09 2.725861e-09
# 5 B.1.617.1 1.323592e-33 1.092436e-31 NA -2.127899e-31 2.154370e-31
# 6 AY.30 8.187325e-02 2.052614e-02 NA 4.164276e-02 1.221038e-01
# 7 other 7.729840e-02 2.593509e-02 NA 2.646655e-02 1.281302e-01
# PLOT MULTINOMIAL FIT
# extrapolate = 30
date.from = as.numeric(as.Date("2021-01-01"))
date.to = as.numeric(as.Date("2021-20-14")) # max(GISAID_sel$DATE_NUM)+extrapolate
# multinomial model predictions (fastest, but no confidence intervals)
predgrid = expand.grid(list(DATE_NUM=seq(date.from, date.to)))
fit_thailand_multi_preds = data.frame(predgrid, as.data.frame(predict(fit2_thailand_multi, newdata=predgrid, type="prob")),check.names=F)
library(tidyr)
library(tidyselect)
fit_thailand_multi_preds = gather(fit_thailand_multi_preds, LINEAGE, prob, all_of(levels_LINEAGE), factor_key=TRUE)
fit_thailand_multi_preds$collection_date = as.Date(fit_thailand_multi_preds$DATE_NUM, origin="1970-01-01")
fit_thailand_multi_preds$LINEAGE = factor(fit_thailand_multi_preds$LINEAGE, levels=levels_LINEAGE)
muller_thailand_mfit = ggplot(data=fit_thailand_multi_preds,
aes(x=collection_date, y=prob, group=LINEAGE)) +
# facet_wrap(~ STATE) +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE, group=LINEAGE), position="stack") +
scale_fill_manual("", values=lineage_cols2) +
annotate("rect", xmin=max(GISAID_sel$DATE_NUM)+1,
xmax=as.Date(date.to, origin="1970-01-01"), ymin=0, ymax=1, alpha=0.4, fill="white") + # extrapolated part
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01","2021-10-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01","2021-10-01"))),1,1),
limits=as.Date(c("2021-01-01",NA)), expand=c(0,0)) +
# guides(color = guide_legend(reverse=F, nrow=1, byrow=T), fill = guide_legend(reverse=F, nrow=1, byrow=T)) +
theme_hc() + theme(legend.position="right",
axis.title.x=element_blank()) +
ylab("Share") +
ggtitle("SPREAD OF SARS-CoV2 VARIANTS OF CONCERN IN THAILAND\n(GISAID data, multinomial fit)")
muller_thailand_mfit
ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_muller plots_multinom fit.png"), width=10, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_muller plots_multinom fit.pdf"), width=10, height=6)
library(ggpubr)
ggarrange(muller_thailand_raw2 + coord_cartesian(xlim=c(as.Date("2021-01-01"),as.Date(date.to, origin="1970-01-01")))+
theme(legend.background = element_rect(fill = alpha("white", 0)),
legend.key = element_rect(fill = alpha("white", 0)),
legend.text=element_text(color = "white")) +
guides(colour = guide_legend(override.aes = list(alpha = 0)),
fill = guide_legend(override.aes = list(alpha = 0))),
muller_thailand_mfit+ggtitle("Multinomial fit"), ncol=1)
ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_muller plots multipanel_multinom fit.png"), width=10, height=10)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_muller plots multipanel_multinom fit.pdf"), width=10, height=10)
# PLOT MODEL FIT WITH DATA & CONFIDENCE INTERVALS
# multinomial model predictions with confidence intervals (but slower)
fit_thailand_multi_preds_withCI = data.frame(emmeans(fit2_thailand_multi,
~ LINEAGE,
by=c("DATE_NUM"),
at=list(DATE_NUM=seq(date.from, date.to, by=1)), # by=XX to speed up things a bit
mode="prob", df=NA))
fit_thailand_multi_preds_withCI$collection_date = as.Date(fit_thailand_multi_preds_withCI$DATE_NUM, origin="1970-01-01")
fit_thailand_multi_preds_withCI$LINEAGE = factor(fit_thailand_multi_preds_withCI$LINEAGE, levels=levels_LINEAGE)
fit_thailand_multi_preds2 = fit_thailand_multi_preds_withCI
# on logit scale:
ymin = 0.001
ymax = 0.999
fit_thailand_multi_preds2$asymp.LCL[fit_thailand_multi_preds2$asymp.LCL<ymin] = ymin
fit_thailand_multi_preds2$asymp.UCL[fit_thailand_multi_preds2$asymp.UCL<ymin] = ymin
fit_thailand_multi_preds2$asymp.UCL[fit_thailand_multi_preds2$asymp.UCL>ymax] = ymax
fit_thailand_multi_preds2$prob[fit_thailand_multi_preds2$prob<ymin] = ymin
plot_thailand_mfit_logit = qplot(data=fit_thailand_multi_preds2, x=collection_date, y=prob, geom="blank") +
# facet_wrap(~ STATE) +
geom_ribbon(aes(y=prob, ymin=asymp.LCL, ymax=asymp.UCL, colour=NULL,
fill=LINEAGE
), alpha=I(0.3)) +
geom_line(aes(y=prob,
colour=LINEAGE
), alpha=I(1)) +
ylab("Share (%)") +
theme_hc() + xlab("") +
ggtitle("SPREAD OF SARS-CoV2 VARIANTS OF CONCERN IN THAILAND\n(GISAID data, multinomial fit)") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01"))),1,1),
limits=as.Date(c("2021-01-01",NA)), expand=c(0,0)) +
scale_y_continuous( trans="logit", breaks=c(10^seq(-5,0),0.5,0.9,0.99,0.999),
labels = c("0.001","0.01","0.1","1","10","100","50","90","99","99.9")) +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
geom_point(data=data_agbyweek,
aes(x=collection_date, y=prop, size=total,
colour=LINEAGE
),
alpha=I(1)) +
scale_size_continuous("total number\nsequenced", trans="identity",
range=c(0.1, 4), limits=c(1,max(data_agbyweek$total)), breaks=c(10,100,1000,10000)) +
# guides(fill=FALSE) +
# guides(colour=FALSE) +
theme(legend.position = "right") +
xlab("Collection date")+
coord_cartesian(xlim=c(as.Date("2021-01-01"),as.Date(date.to, origin="1970-01-01")), ylim=c(0.001, 0.9901), expand=c(0,0))
plot_thailand_mfit_logit
ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_multinom fit_logit scale.png"), width=10, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_multinom fit_logit scale.pdf"), width=10, height=6)
# on response scale:
plot_thailand_mfit = qplot(data=fit_thailand_multi_preds2, x=collection_date, y=100*prob, geom="blank") +
# facet_wrap(~ STATE) +
geom_ribbon(aes(y=100*prob, ymin=100*asymp.LCL, ymax=100*asymp.UCL, colour=NULL,
fill=LINEAGE
), alpha=I(0.3)) +
geom_line(aes(y=100*prob,
colour=LINEAGE
), alpha=I(1)) +
ylab("Share (%)") +
theme_hc() + xlab("") +
ggtitle("SPREAD OF SARS-CoV2 VARIANTS OF CONCERN IN THAILAND\n(GISAID data, multinomial fit)") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01","2021-10-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01","2021-10-01"))),1,1),
limits=as.Date(c("2021-01-01",NA)), expand=c(0,0)) +
# scale_y_continuous( trans="logit", breaks=c(10^seq(-5,0),0.5,0.9,0.99,0.999),
# labels = c("0.001","0.01","0.1","1","10","100","50","90","99","99.9")) +
coord_cartesian(xlim=as.Date(c("2021-01-01",NA)),
ylim=c(0,100), expand=c(0,0)) +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
geom_point(data=data_agbyweek,
aes(x=collection_date, y=100*prop, size=total,
colour=LINEAGE
),
alpha=I(1)) +
scale_size_continuous("total number\nsequenced", trans="identity",
range=c(0.5, 5), limits=c(1,max(data_agbyweek$total)), breaks=c(100,1000,10000)) +
# guides(fill=FALSE) +
# guides(colour=FALSE) +
theme(legend.position = "right") +
xlab("Collection date")
plot_thailand_mfit
ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_multinom fit_response scale.png"), width=10, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\thailand_multinom fit_response scale.pdf"), width=10, height=6)
# PLOTS OF NEW CASES PER DAY BY VARIANT & EFFECTIVE REPRODUCTION NUMBER BY VARIANT THROUGH TIME ####
# load case data
library(covidregionaldata)
library(dplyr)
library(ggplot2)
library(scales)
cases_tot = as.data.frame(get_national_data(countries = "Thailand"))
cases_tot = cases_tot[cases_tot$date>=as.Date("2020-08-01"),]
cases_tot$DATE_NUM = as.numeric(cases_tot$date)
# cases_tot$BANKHOLIDAY = bankholiday(cases_tot$date)
cases_tot$WEEKDAY = weekdays(cases_tot$date)
cases_tot = cases_tot[cases_tot$date<=(max(cases_tot$date)-3),] # cut off data from last 3 days (incomplete)
range(cases_tot$date)
# smooth out weekday effects in case nrs using GAM (if testing data is available one could correct for testing intensity as well)
library(mgcv)
k=25
fit_cases = gam(cases_new ~ s(DATE_NUM, bs="cs", k=k, m=c(2), fx=F) +
WEEKDAY, # +
# BANKHOLIDAY,
# s(TESTS_ALL, bs="cs", k=8, fx=F),
family=poisson(log), data=cases_tot,
method = "REML",
knots = list(DATE_NUM = c(min(cases_tot$DATE_NUM)-14,
seq(min(cases_tot$DATE_NUM)+0.7*diff(range(cases_tot$DATE_NUM))/(k-2),
max(cases_tot$DATE_NUM)-0.7*diff(range(cases_tot$DATE_NUM))/(k-2), length.out=k-2),
max(cases_tot$DATE_NUM)+14))
)
BIC(fit_cases)
# STACKED AREA CHART OF NEW CASES BY VARIANT (MULTINOMIAL FIT MAPPED ONTO CASE DATA) ####
fit_thailand_multi_preds_withCI$totcases = cases_tot$cases_new[match(round(fit_thailand_multi_preds_withCI$DATE_NUM),cases_tot$DATE_NUM)]
fit_thailand_multi_preds_withCI$cases = fit_thailand_multi_preds_withCI$totcases * fit_thailand_multi_preds_withCI$prob
fit_thailand_multi_preds_withCI$cases[fit_thailand_multi_preds_withCI$cases<=0.001] = NA
cases_emmeans = as.data.frame(emmeans(fit_cases, ~ DATE_NUM, at=list(DATE_NUM=seq(date.from, date.to, by=0.5), BANHOLIDAY="no"), type="response"))
fit_thailand_multi_preds_withCI$smoothed_totcases = cases_emmeans$rate[match(fit_thailand_multi_preds_withCI$DATE_NUM,cases_emmeans$DATE_NUM)]
fit_thailand_multi_preds_withCI$smoothed_cases = fit_thailand_multi_preds_withCI$smoothed_totcases * fit_thailand_multi_preds_withCI$prob
fit_thailand_multi_preds_withCI$smoothed_cases[fit_thailand_multi_preds_withCI$smoothed_cases<=0.001] = NA
fit_thailand_multi_preds_withCI$LINEAGE = factor(fit_thailand_multi_preds_withCI$LINEAGE, levels=levels_LINEAGE)
ggplot(data=fit_thailand_multi_preds_withCI[fit_thailand_multi_preds_withCI$collection_date>=as.Date("2021-01-01"),],
aes(x=collection_date, y=cases, group=LINEAGE)) +
# facet_wrap(~ REGION, scale="free", ncol=3) +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE, group=LINEAGE), position="stack") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01"))),1,1),
# limits=c(as.Date("2021-03-01"),max(cases_tot$date)),
expand=c(0,0)) +
# guides(color = guide_legend(reverse=F, nrow=1, byrow=T), fill = guide_legend(reverse=F, nrow=1, byrow=T)) +
theme_hc() + theme(legend.position="right") +
ylab("New confirmed cases per day") + xlab("Date of diagnosis") +
ggtitle("NEW CONFIRMED SARS-CoV2 CASES PER DAY BY VARIANT\nIN THAILAND\n(case data & multinomial fit to GISAID data)") +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
coord_cartesian(xlim=c(as.Date("2021-01-01"),NA))
ggsave(file=paste0(".\\plots\\",plotdir,"\\cases per day_stacked area multinomial fit raw case data.png"), width=8, height=6)
ggplot(data=fit_thailand_multi_preds_withCI[fit_thailand_multi_preds_withCI$collection_date>=as.Date("2021-01-01")&
fit_thailand_multi_preds_withCI$collection_date<=max(cases_tot$date),],
aes(x=collection_date-7, y=smoothed_cases, group=LINEAGE)) +
# facet_wrap(~ REGION, scale="free", ncol=3) +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE, group=LINEAGE), position="stack") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01"))),1,1),
# limits=c(as.Date("2021-03-01"),today),
expand=c(0,0)) +
# guides(color = guide_legend(reverse=F, nrow=1, byrow=T), fill = guide_legend(reverse=F, nrow=1, byrow=T)) +
theme_hc() + theme(legend.position="right") +
ylab("New confirmed cases per day (smoothed)") + xlab("Date of infection") +
ggtitle("NEW CONFIRMED SARS-CoV2 CASES PER DAY BY VARIANT\nIN THAILAND\n(case data & multinomial fit to GISAID data)") +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
coord_cartesian(xlim=c(as.Date("2021-01-01"),max(cases_tot$date)))
ggsave(file=paste0(".\\plots\\",plotdir,"\\cases per day_smoothed_stacked area multinomial fit case data.png"), width=8, height=6)
# EFFECTIVE REPRODUCTION NUMBER BY VARIANT THROUGH TIME ####
# Function to calculate Re values from intrinsic growth rate
# cf. https://github.com/epiforecasts/EpiNow2/blob/5015e75f7048c2580b2ebe83e46d63124d014861/R/utilities.R#L109
# https://royalsocietypublishing.org/doi/10.1098/rsif.2020.0144
# (assuming gamma distributed gen time)
Re.from.r <- function(r, gamma_mean=4.7, gamma_sd=2.9) { # Nishiura et al. 2020, or use values from Ferretti et al. 2020 (gamma_mean=5.5, gamma_sd=1.8)
k <- (gamma_sd / gamma_mean)^2
R <- (1 + k * r * gamma_mean)^(1 / k)
return(R)
}
# calculate average instantaneous growth rates & 95% CLs using emtrends ####
# based on the slope of the GAM fit on a log link scale
avg_r_cases = as.data.frame(emtrends(fit_cases, ~ DATE_NUM, var="DATE_NUM",
at=list(DATE_NUM=seq(date.from,
date.to)#,
# BANKHOLIDAY="no"
), # weekday="Wednesday",
type="link"))
colnames(avg_r_cases)[2] = "r"
colnames(avg_r_cases)[5] = "r_LOWER"
colnames(avg_r_cases)[6] = "r_UPPER"
avg_r_cases$DATE = as.Date(avg_r_cases$DATE_NUM, origin="1970-01-01") # -7 TO CALCULATE BACK TO INFECTION DATE
avg_r_cases$Re = Re.from.r(avg_r_cases$r)
avg_r_cases$Re_LOWER = Re.from.r(avg_r_cases$r_LOWER)
avg_r_cases$Re_UPPER = Re.from.r(avg_r_cases$r_UPPER)
avg_r_cases = avg_r_cases[complete.cases(avg_r_cases),]
qplot(data=avg_r_cases, x=DATE-7, y=Re, ymin=Re_LOWER, ymax=Re_UPPER, geom="ribbon", alpha=I(0.5), fill=I("steelblue")) +
# facet_wrap(~ REGION) +
geom_line() + theme_hc() + xlab("Date of infection") +
scale_x_continuous(breaks=as.Date(c("2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01")),
labels=c("M","A","M","J","J","A","S","O","N","D","J","F","M","A","M","J","J","A","S")) +
# scale_y_continuous(limits=c(1/2, 2), trans="log2") +
geom_hline(yintercept=1, colour=I("red")) +
ggtitle("Re IN THAILAND AT MOMENT OF INFECTION BASED ON NEW CASES") +
# labs(tag = tag) +
# theme(plot.margin = margin(t = 20, r = 10, b = 20, l = 0)) +
theme(plot.tag.position = "bottomright",
plot.tag = element_text(vjust = 1, hjust = 1, size=8)) # +
# coord_cartesian(xlim=c(as.Date("2020-01-01"),NA))
# calculate above-average intrinsic growth rates per day of each variant over time based on multinomial fit using emtrends weighted effect contrasts ####
# for best model fit3_sanger_multi
above_avg_r_variants0 = do.call(rbind, lapply(seq(date.from,
date.to),
function (d) {
wt = as.data.frame(emmeans(fit2_thailand_multi, ~ LINEAGE , at=list(DATE_NUM=d), type="response"))$prob # important: these should sum to 1
# wt = rep(1/length(levels_VARIANTS), length(levels_VARIANTS)) # this would give equal weights, equivalent to emmeans:::eff.emmc(levs=levels_LINEAGE)
cons = lapply(seq_along(wt), function (i) { con = -wt; con[i] = 1 + con[i]; con })
names(cons) = seq_along(cons)
EMT = emtrends(fit2_thailand_multi, ~ LINEAGE , by=c("DATE_NUM"),
var="DATE_NUM", mode="latent",
at=list(DATE_NUM=d))
out = as.data.frame(confint(contrast(EMT, cons), adjust="none", df=NA))
# sum(out$estimate*wt) # should sum to zero
return(out) } ))
above_avg_r_variants = above_avg_r_variants0
above_avg_r_variants$contrast = factor(above_avg_r_variants$contrast,
levels=1:length(levels_LINEAGE),
labels=levels_LINEAGE)
above_avg_r_variants$variant = above_avg_r_variants$contrast # gsub(" effect|\\(|\\)","",above_avg_r_variants$contrast)
above_avg_r_variants$collection_date = as.Date(above_avg_r_variants$DATE_NUM, origin="1970-01-01")
range(above_avg_r_variants$collection_date) # "2021-01-04" "2021-07-30"
above_avg_r_variants$avg_r = avg_r_cases$r[match(above_avg_r_variants$collection_date,
avg_r_cases$DATE)] # average growth rate of all lineages calculated from case nrs
above_avg_r_variants$r = above_avg_r_variants$avg_r+above_avg_r_variants$estimate
above_avg_r_variants$r_LOWER = above_avg_r_variants$avg_r+above_avg_r_variants$asymp.LCL
above_avg_r_variants$r_UPPER = above_avg_r_variants$avg_r+above_avg_r_variants$asymp.UCL
above_avg_r_variants$Re = Re.from.r(above_avg_r_variants$r)
above_avg_r_variants$Re_LOWER = Re.from.r(above_avg_r_variants$r_LOWER)
above_avg_r_variants$Re_UPPER = Re.from.r(above_avg_r_variants$r_UPPER)
df = data.frame(contrast=NA,
DATE_NUM=avg_r_cases$DATE_NUM, # -7 to calculate back to time of infection
# REGION=avg_r_cases$REGION,
estimate=NA,
SE=NA,
df=NA,
asymp.LCL=NA,
asymp.UCL=NA,
# p.value=NA,
collection_date=avg_r_cases$DATE,
variant="avg",
avg_r=avg_r_cases$r,
r=avg_r_cases$r,
r_LOWER=avg_r_cases$r_LOWER,
r_UPPER=avg_r_cases$r_UPPER,
Re=avg_r_cases$Re,
Re_LOWER=avg_r_cases$Re_LOWER,
Re_UPPER=avg_r_cases$Re_UPPER)
# df = df[df$DATE_NUM<=max(above_avg_r_variants$DATE_NUM)&df$DATE_NUM>=(min(above_avg_r_variants$DATE_NUM)+7),]
above_avg_r_variants = rbind(above_avg_r_variants, df)
above_avg_r_variants$variant = factor(above_avg_r_variants$variant, levels=c(levels_LINEAGE,"avg"))
above_avg_r_variants$prob = fit_thailand_multi_preds_withCI$prob[match(interaction(above_avg_r_variants$DATE_NUM,
above_avg_r_variants$variant),
interaction(fit_thailand_multi_preds_withCI$DATE_NUM,
fit_thailand_multi_preds_withCI$LINEAGE))]
above_avg_r_variants2 = above_avg_r_variants
ymax = 3
ymin = 1/3
above_avg_r_variants2$Re[above_avg_r_variants2$Re>=ymax] = NA
above_avg_r_variants2$Re[above_avg_r_variants2$Re<=ymin] = NA
above_avg_r_variants2$Re_LOWER[above_avg_r_variants2$Re_LOWER>=ymax] = ymax
above_avg_r_variants2$Re_LOWER[above_avg_r_variants2$Re_LOWER<=ymin] = ymin
above_avg_r_variants2$Re_UPPER[above_avg_r_variants2$Re_UPPER>=ymax] = ymax
above_avg_r_variants2$Re_UPPER[above_avg_r_variants2$Re_UPPER<=ymin] = ymin
above_avg_r_variants2$Re[above_avg_r_variants2$prob<0.01] = NA
above_avg_r_variants2$Re_LOWER[above_avg_r_variants2$prob<0.01] = NA
above_avg_r_variants2$Re_UPPER[above_avg_r_variants2$prob<0.01] = NA
qplot(data=above_avg_r_variants2[!((above_avg_r_variants2$variant %in% c("other"))|above_avg_r_variants2$collection_date>max(cases_tot$DATE)),],
x=collection_date-7, # -7 to calculate back to date of infection
y=Re, ymin=Re_LOWER, ymax=Re_UPPER, geom="ribbon", colour=variant, fill=variant, alpha=I(0.5),
group=variant, linetype=I(0)) +
# facet_wrap(~ REGION) +
# geom_ribbon(aes(fill=variant, colour=variant), alpha=I(0.5))
geom_line(aes(colour=variant), lwd=I(0.72)) + theme_hc() + xlab("Date of infection") +
scale_x_continuous(breaks=as.Date(c("2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01","2021-08-01","2021-09-01")),
labels=c("M","A","M","J","J","A","S","O","N","D","J","F","M","A","M","J","J","A","S")) +
# scale_y_continuous(limits=c(1/ymax,ymax), trans="log2") +
geom_hline(yintercept=1, colour=I("red")) +
ggtitle("Re VALUES OF SARS-CoV2 VARIANTS IN THAILAND\nAT MOMENT OF INFECTION\n(based on case data & multinomial fit to GISAID data)") +
# labs(tag = tag) +
# theme(plot.margin = margin(t = 20, r = 10, b = 20, l = 0)) +
theme(plot.tag.position = "bottomright",
plot.tag = element_text(vjust = 1, hjust = 1, size=8)) +
coord_cartesian(xlim=c(as.Date("2021-01-01"),max(cases_tot$date))) +
scale_fill_manual("variant", values=c(head(lineage_cols2,-1),"black")) +
scale_colour_manual("variant", values=c(head(lineage_cols2,-1),"black")) +
theme(legend.position="right")
ggsave(file=paste0(".\\plots\\",plotdir,"\\Re values per variant_avgRe_from_cases_with clipping.png"), width=8, height=6)