forked from GuillaumeDerval/newcovid_belgium
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathanalysis_VOCs_GISAID genomic epidemiology metadata_Portugal.R
720 lines (622 loc) · 42.5 KB
/
analysis_VOCs_GISAID genomic epidemiology metadata_Portugal.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
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
# ANALYSIS OF GROWTH ADVANTAGE OF DIFFERENT VOCs IN PORTUGAL (GISAID GENOMIC EPIDEMIOLOGY METADATA)
# T. Wenseleers
# last update 7 July 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)
today = as.Date(Sys.time()) # we use the file date version as our definition of "today"
today = as.Date("2021-07-07")
today_num = as.numeric(today)
plotdir = "Portugal_GISAID"
suppressWarnings(dir.create(paste0(".//plots//",plotdir)))
# import GISAID genomic epidemiology metadata (file version metadata_2021-07-05_12-45.tsv.gz)
GISAID = read_tsv(gzfile(".//data//GISAID_genomic_epidemiology//metadata_2021-07-05_12-45.tsv.gz"), col_types = cols(.default = "c"))
GISAID = as.data.frame(GISAID)
GISAID$date = as.Date(GISAID$date)
GISAID = GISAID[!is.na(GISAID$date),]
unique(GISAID$host)
# [1] "Human" "Environment" "Feline" "unknown"
# [5] "Rhinolophus shameli" "Rhinolophus malayanus" "Rhinolophus pusillus" "Rhinolophus sinicus"
# [9] "Rhinolophus stheno" "Rhinolophus affinis" "Felis catus" "Canis lupus familiaris"
# [13] "Gorilla gorilla gorilla" "Mesocricetus auratus" "Prionailurus bengalensis euptilurus" "Panthera leo"
# [17] "Mink" "Mustela putorius furo" "Chlorocebus sabaeus" "Mus musculus"
# [21] "Mus musculus (BALB/c mice)" "Manis javanica" "Manis pentadactyla" "Panthera tigris jacksoni"
GISAID[GISAID$host!="Human","strain"]
GISAID = GISAID[GISAID$host=="Human",]
GISAID = GISAID[GISAID$date>=as.Date("2020-01-01"),]
range(GISAID$date) # "2020-01-01" "2021-07-01"
firstdetB16172 = GISAID[GISAID$pango_lineage=="B.1.617.2",]
firstdetB16172 = firstdetB16172[!is.na(firstdetB16172$date),]
firstdetB16172 = firstdetB16172[firstdetB16172$date==min(firstdetB16172$date),]
firstdetB16172 # 7 sept 63r old male from Madhya Pradesh
# GISAID = GISAID[grepl("2021-", GISAID$date),]
sum(is.na(GISAID$purpose_of_sequencing)) == nrow(GISAID) # field purpose_of_sequencing left blank unfortunately
nrow(GISAID) # 1901769
GISAID$Week = lubridate::week(GISAID$date)
GISAID$Year = lubridate::year(GISAID$date)
GISAID$Year_Week = interaction(GISAID$Year,GISAID$Week)
library(lubridate)
GISAID$floor_date = as.Date(as.character(cut(GISAID$date, "week")))+3.5 # week midpoint date
GISAID$DATE_NUM = as.numeric(GISAID$date)
colnames(GISAID)
unique(GISAID$region)
# "Asia" "Europe" "Africa" "South America" "Oceania" "North America"
unique(GISAID$country)
unique(GISAID$division) # = city or province or region, sometimes just country
unique(GISAID$location) # = city
length(unique(GISAID$country[grepl("B.1.617",GISAID$pango_lineage,fixed=T)])) # B.1.617+ now found in 67 countries
table(GISAID$pango_lineage[grepl("B.1.617",GISAID$pango_lineage,fixed=T)])
# B.1.617 B.1.617.1 B.1.617.2 B.1.617.3
# 2 4388 51068 147
GISAID$pango_lineage[grepl("B.1.177",GISAID$pango_lineage,fixed=T)] = "B.1.177+"
GISAID$pango_lineage[grepl("B.1.36\\>",GISAID$pango_lineage)] = "B.1.36+"
sel_target_VOC = "B.1.617"
GISAID$LINEAGE1 = GISAID$pango_lineage
GISAID$LINEAGE2 = GISAID$pango_lineage
GISAID[grepl(sel_target_VOC, GISAID$LINEAGE1, fixed=TRUE),"LINEAGE1"] = paste0(sel_target_VOC,"+") # in LINEAGE1 we recode B.1.617.1,2&3 all as B.1.617+
table_country_lineage = as.data.frame(table(GISAID$country, GISAID$LINEAGE1))
colnames(table_country_lineage) = c("Country","Lineage","Count")
tblB1617 = table_country_lineage[grepl(sel_target_VOC, table_country_lineage$Lineage, fixed=T)&table_country_lineage$Count>10,]
tblB1617
# Country Lineage Count
# 170497 Australia B.1.617+ 252
# 170501 Bahrain B.1.617+ 24
# 170502 Bangladesh B.1.617+ 44
# 170505 Belgium B.1.617+ 247
# 170513 Brazil B.1.617+ 16
# 170519 Canada B.1.617+ 346
# 170530 Czech Republic B.1.617+ 17
# 170532 Denmark B.1.617+ 121
# 170542 Finland B.1.617+ 19
# 170543 France B.1.617+ 141
# 170548 Germany B.1.617+ 836
# 170562 India B.1.617+ 7435
# 170563 Indonesia B.1.617+ 75
# 170564 Iran B.1.617+ 11
# 170566 Ireland B.1.617+ 299
# 170567 Israel B.1.617+ 63
# 170568 Italy B.1.617+ 184
# 170570 Japan B.1.617+ 170
# 170581 Luxembourg B.1.617+ 58
# 170583 Malawi B.1.617+ 26
# 170584 Malaysia B.1.617+ 12
# 170588 Mexico B.1.617+ 48
# 170596 Nepal B.1.617+ 34
# 170597 Netherlands B.1.617+ 85
# 170598 New Zealand B.1.617+ 17
# 170602 Norway B.1.617+ 69
# 170611 Poland B.1.617+ 71
# 170612 Portugal B.1.617+ 126
# 170613 Qatar B.1.617+ 23
# 170615 Romania B.1.617+ 19
# 170616 Russia B.1.617+ 278
# 170627 Singapore B.1.617+ 762
# 170632 South Africa B.1.617+ 21
# 170633 South Korea B.1.617+ 32
# 170635 Spain B.1.617+ 264
# 170638 Sweden B.1.617+ 42
# 170639 Switzerland B.1.617+ 113
# 170641 Thailand B.1.617+ 94
# 170651 United Kingdom B.1.617+ 40092
# 170653 USA B.1.617+ 2859
# 170656 Vietnam B.1.617+ 54
sel_countries_target = unique(as.character(table_country_lineage[grepl(sel_target_VOC, table_country_lineage$Lineage)&table_country_lineage$Count>100,]$Country))
sel_countries_target
# [1] "Australia" "Belgium" "Canada" "Denmark" "France" "Germany" "India" "Ireland"
# [9] "Italy" "Japan" "Portugal" "Russia" "Singapore" "Spain" "Switzerland" "United Kingdom"
# [17] "USA"
sel_ref_lineage = "B.1.1.7"
sel_countries_ref = as.character(table_country_lineage[table_country_lineage$Lineage==sel_ref_lineage&table_country_lineage$Count>10&table_country_lineage$Country %in% sel_countries_target,]$Country)
sel_countries_ref
# [1] "Australia" "Belgium" "Canada" "Denmark" "France" "Germany" "India" "Ireland"
# [9] "Italy" "Japan" "Portugal" "Russia" "Singapore" "Spain" "Switzerland" "United Kingdom"
# [17] "USA"
sel_countries = intersect(sel_countries_target, sel_countries_ref)
sel_countries
# [1] "Australia" "Belgium" "Canada" "Denmark" "France" "Germany" "India" "Ireland"
# [9] "Italy" "Japan" "Portugal" "Russia" "Singapore" "Spain" "Switzerland" "United Kingdom"
# [17] "USA"
# sel_countries = sel_countries[!(sel_countries %in% c("Japan","USA"))] # Japan is almost only import & for USA we do separate analysis by state
# ANALYSIS OF VOCs IN PORTUGAL ####
sel_countries = "Portugal"
tblB117 = table_country_lineage[table_country_lineage$Lineage==sel_ref_lineage&table_country_lineage$Count>10&table_country_lineage$Country %in% sel_countries,]
tblB117
GISAID_sel = GISAID[GISAID$country %in% sel_countries,]
# use data from Feb 1 onwards
GISAID_sel = GISAID_sel[GISAID_sel$date>=as.Date("2021-02-01"),]
nrow(GISAID_sel) # 6182
range(GISAID_sel$date) # "2021-02-01" "2021-06-23"
nrow(GISAID_sel) # 6182
unique(GISAID_sel$country)
rowSums(table(GISAID_sel$LINEAGE1,GISAID_sel$country))
# GISAID_sel = GISAID_sel[GISAID_sel$country_exposure=="India"&GISAID_sel$country!="India",]
# nrow(GISAID_sel[is.na(GISAID_sel$LINEAGE1),]) # 0 unknown pango clade
GISAID_sel = GISAID_sel[!is.na(GISAID_sel$LINEAGE1),]
nrow(GISAID_sel) # 6182
GISAID_sel = GISAID_sel[GISAID_sel$country==GISAID_sel$country_exposure,] # we remove travel-related cases (none indicated here as such)
nrow(GISAID_sel) # 6182
sum(GISAID_sel$LINEAGE1=="B.1.617+") # 126
unique(GISAID_sel$country[GISAID_sel$LINEAGE1=="B.1.1.7"])
sum(GISAID_sel$LINEAGE1=="B.1.1.7") # 4206
sum(GISAID_sel$LINEAGE1=="B.1.1.519") # 0
table(GISAID_sel$LINEAGE1)
table(GISAID_sel$LINEAGE2)
main_lineages = names(table(GISAID_sel$LINEAGE1))[100*table(GISAID_sel$LINEAGE1)/sum(table(GISAID_sel$LINEAGE1)) > 3]
main_lineages
# "B.1.1.7" "B.1.177+" "B.1.617+" "other"
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")
main_lineages = union(main_lineages, VOCs)
GISAID_sel$LINEAGE1[!(GISAID_sel$LINEAGE1 %in% main_lineages)] = "other" # minority lineages & non-VOCs
GISAID_sel$LINEAGE2[!(GISAID_sel$LINEAGE2 %in% main_lineages)] = "other" # minority lineages & non-VOCs
remove1 = names(table(GISAID_sel$LINEAGE1))[table(GISAID_sel$LINEAGE1)/sum(table(GISAID_sel$LINEAGE1)) < 0.01]
remove1 = remove1[!(remove1 %in% c("B.1.351","B.1.1.7","P.1","B.1.617+","B.1.1.519"))]
remove2 = names(table(GISAID_sel$LINEAGE2))[table(GISAID_sel$LINEAGE2)/sum(table(GISAID_sel$LINEAGE2)) < 0.01]
remove2 = remove2[!(remove2 %in% c("B.1.351","B.1.1.7","P.1","B.1.617.2","B.1.617.1","B.1.1.519"))]
GISAID_sel$LINEAGE1[(GISAID_sel$LINEAGE1 %in% remove1)] = "other" # minority VOCs
GISAID_sel$LINEAGE2[(GISAID_sel$LINEAGE2 %in% remove2)] = "other" # minority VOCs
table(GISAID_sel$LINEAGE1)
# B.1 B.1.1 B.1.1.7 B.1.177+ B.1.351 B.1.617+ other P.1
# 454 802 4679 1341 96 728 1603 169
GISAID_sel$LINEAGE1 = factor(GISAID_sel$LINEAGE1)
GISAID_sel$LINEAGE1 = relevel(GISAID_sel$LINEAGE1, ref="B.1.1.7") # we code UK strain as the reference level
levels(GISAID_sel$LINEAGE1)
# "B.1.1.7" "B.1" "B.1.1" "B.1.160" "B.1.177+" "B.1.351" "B.1.617+" "B.1.91" "other" "P.1"
levels_LINEAGE1 = c("B.1.1.7","B.1.177+",
"B.1.351","P.1","B.1.617+","other")
GISAID_sel$LINEAGE1 = factor(GISAID_sel$LINEAGE1, levels=levels_LINEAGE1)
GISAID_sel$LINEAGE2 = factor(GISAID_sel$LINEAGE2)
GISAID_sel$LINEAGE2 = relevel(GISAID_sel$LINEAGE2, ref="B.1.1.7") # we code UK strain as the reference level
levels(GISAID_sel$LINEAGE2)
# "B.1.1.7" "B.1.177+" "B.1.351" "B.1.617.1" "B.1.617.2" "other" "P.1"
levels_LINEAGE2 = c("B.1.1.7","B.1.177+",
"B.1.351","P.1","B.1.617.1","B.1.617.2","other")
GISAID_sel$LINEAGE2 = factor(GISAID_sel$LINEAGE2, levels=levels_LINEAGE2)
# GISAID_sel = GISAID_sel[GISAID_sel$division!="India",]
table(GISAID_sel$country)
# B.1.617+ cases before Apr 14 are likely mostly imported cases, so we remove those
GISAID_sel = GISAID_sel[-which(grepl("B.1.617", GISAID_sel$pango_lineage, fixed=TRUE)&GISAID_sel$date<=as.Date("2021-04-14")),]
table(GISAID_sel$LINEAGE2)
# B.1.1.7 B.1.177+ B.1.351 P.1 B.1.617.1 B.1.617.2 other
# 4387 355 93 169 1 719 450
range(GISAID_sel$date) # "2020-03-01" "2021-06-02"
# aggregated data to make Muller plots of raw data
# aggregate by day to identify days on which INSA performed baseline surveillance (days with a lot of sequences)
# we subset the data to just those days to avoid sampling biases (delta infection clusters etc)
data_agbyday2 = as.data.frame(table(GISAID_sel$date, GISAID_sel$LINEAGE2))
colnames(data_agbyday2) = c("date", "LINEAGE2", "count")
data_agbyday2_sum = aggregate(count ~ date, data=data_agbyday2, sum)
data_agbyday2$total = data_agbyday2_sum$count[match(data_agbyday2$date, data_agbyday2_sum$date)]
sum(data_agbyday2[data_agbyday2$LINEAGE2=="B.1.617.1","total"]) == nrow(GISAID_sel) # correct
data_agbyday2$date = as.Date(as.character(data_agbyday2$date))
data_agbyday2$LINEAGE2 = factor(data_agbyday2$LINEAGE2, levels=levels_LINEAGE2)
data_agbyday2$date_num = as.numeric(data_agbyday2$date)
data_agbyday2$prop = data_agbyday2$count/data_agbyday2$total
data_agbyday2$floor_date = NULL
qplot(data=data_agbyday2, x=date, y=total, colour=total>20, fill=total>20, geom="col")
GISAID_sel$total_sequenced_on_that_day = data_agbyday2$total[match(GISAID_sel$date, data_agbyday2$date)]
GISAID_sel = GISAID_sel[GISAID_sel$total_sequenced_on_that_day>20,] # dates on which baseline surveillance was performed
nrow(GISAID_sel) # 5710
# aggregated by week for selected variant lineages
data_agbyweek1 = as.data.frame(table(GISAID_sel$floor_date, GISAID_sel$LINEAGE1))
colnames(data_agbyweek1) = c("floor_date", "LINEAGE1", "count")
data_agbyweek1_sum = aggregate(count ~ floor_date, data=data_agbyweek1, sum)
data_agbyweek1$total = data_agbyweek1_sum$count[match(data_agbyweek1$floor_date, data_agbyweek1_sum$floor_date)]
sum(data_agbyweek1[data_agbyweek1$LINEAGE1=="B.1.617+","total"]) == nrow(GISAID_sel) # correct
data_agbyweek1$collection_date = as.Date(as.character(data_agbyweek1$floor_date))
data_agbyweek1$LINEAGE1 = factor(data_agbyweek1$LINEAGE1, levels=levels_LINEAGE1)
data_agbyweek1$collection_date_num = as.numeric(data_agbyweek1$collection_date)
data_agbyweek1$prop = data_agbyweek1$count/data_agbyweek1$total
data_agbyweek1$floor_date = NULL
data_agbyweek2 = as.data.frame(table(GISAID_sel$floor_date, GISAID_sel$LINEAGE2))
colnames(data_agbyweek2) = c("floor_date", "LINEAGE2", "count")
data_agbyweek2_sum = aggregate(count ~ floor_date, data=data_agbyweek2, sum)
data_agbyweek2$total = data_agbyweek2_sum$count[match(data_agbyweek2$floor_date, data_agbyweek2_sum$floor_date)]
sum(data_agbyweek2[data_agbyweek2$LINEAGE2=="B.1.617.1","total"]) == nrow(GISAID_sel) # correct
data_agbyweek2$collection_date = as.Date(as.character(data_agbyweek2$floor_date))
data_agbyweek2$LINEAGE2 = factor(data_agbyweek2$LINEAGE2, levels=levels_LINEAGE2)
data_agbyweek2$collection_date_num = as.numeric(data_agbyweek2$collection_date)
data_agbyweek2$prop = data_agbyweek2$count/data_agbyweek2$total
data_agbyweek2$floor_date = NULL
# MULLER PLOT (RAW DATA)
library(scales)
n1 = length(levels(GISAID_sel$LINEAGE1))
lineage_cols1 = hcl(h = seq(15, 320, length = n1), l = 65, c = 200)
lineage_cols1[which(levels(GISAID_sel$LINEAGE1)=="B.1.1.7")] = "#0085FF"
lineage_cols1[which(levels(GISAID_sel$LINEAGE1)=="B.1.351")] = "#9A9D00"
lineage_cols1[which(levels(GISAID_sel$LINEAGE1)=="P.1")] = "cyan3"
lineage_cols1[which(levels(GISAID_sel$LINEAGE1)=="B.1.617+")] = "magenta"
lineage_cols1[which(levels(GISAID_sel$LINEAGE1)=="other")] = "grey75"
n2 = length(levels(GISAID_sel$LINEAGE2))
lineage_cols2 = hcl(h = seq(15, 320, length = n2), l = 65, c = 200)
lineage_cols2[which(levels(GISAID_sel$LINEAGE2)=="B.1.1.7")] = "#0085FF"
lineage_cols2[which(levels(GISAID_sel$LINEAGE2)=="B.1.351")] = "#9A9D00"
lineage_cols2[which(levels(GISAID_sel$LINEAGE2)=="P.1")] = "cyan3"
lineage_cols2[which(levels(GISAID_sel$LINEAGE2)=="B.1.617.1")] = muted("magenta")
lineage_cols2[which(levels(GISAID_sel$LINEAGE2)=="B.1.617.2")] = "magenta"
lineage_cols2[which(levels(GISAID_sel$LINEAGE2)=="other")] = "grey75"
muller_portugal_raw2 = ggplot(data=data_agbyweek2, aes(x=collection_date, y=count, group=LINEAGE2)) +
# 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=LINEAGE2, group=LINEAGE2), 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")),
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"))),1,1),
limits=as.Date(c("2021-02-01",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 PORTUGAL\n(GISAID data)")
# +
# coord_cartesian(xlim=c(1,max(GISAID_sel$Week)))
muller_portugal_raw2
ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_muller plots_raw data.png"), width=8, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_muller plots_raw data.pdf"), width=8, height=6)
# multinomial fits
data_agbyweek2$LINEAGE2 = relevel(data_agbyweek2$LINEAGE2, ref="B.1.1.7")
data_agbyweek2$DATE_NUM = as.numeric(data_agbyweek2$collection_date)
library(nnet)
library(splines)
set.seed(1)
fit1_portugal_multi = nnet::multinom(LINEAGE2 ~ scale(DATE_NUM), weights=count, data=data_agbyweek2, maxit=1000)
fit2_portugal_multi = nnet::multinom(LINEAGE2 ~ ns(DATE_NUM, df=2), weights=count, data=data_agbyweek2, maxit=1000)
BIC(fit1_portugal_multi, fit2_portugal_multi)
# df BIC
# fit1_portugal_multi 12 8176.180
# fit2_portugal_multi 18 8043.871
# growth rate advantage compared to UK type B.1.1.7 (difference in growth rate per day)
emtrportugal = emtrends(fit2_portugal_multi, trt.vs.ctrl ~ LINEAGE2,
var="DATE_NUM", mode="latent",
at=list(DATE_NUM=max(GISAID_sel$DATE_NUM)))
delta_r_portugal = data.frame(confint(emtrportugal,
adjust="none", df=NA)$contrasts,
p.value=as.data.frame(emtrportugal$contrasts)$p.value)
delta_r_portugal
# contrast estimate SE df asymp.LCL asymp.UCL p.value
# 1 (B.1.177+) - B.1.1.7 -0.05581888 0.025483171 NA -0.10576498 -0.005872785 1.741393e-01
# 2 B.1.351 - B.1.1.7 -0.03277813 0.014848702 NA -0.06188106 -0.003675212 1.690578e-01
# 3 P.1 - B.1.1.7 -0.00632620 0.007622421 NA -0.02126587 0.008613471 8.598001e-01
# 4 B.1.617.1 - B.1.1.7 -0.98672776 0.599795722 NA -2.16230577 0.188850256 4.044027e-01
# 5 B.1.617.2 - B.1.1.7 0.09301972 0.011259119 NA 0.07095225 0.115087184 8.757008e-07
# 6 other - B.1.1.7 0.04823980 0.004840895 NA 0.03875182 0.057727779 5.382958e-08
# fitted prop of different LINEAGES in the portugal today
# 99% [97%-100%] now estimated to be B.1.617.2 across all regions
multinom_preds_today_avg = data.frame(emmeans(fit2_portugal_multi, ~ LINEAGE2|1,
at=list(DATE_NUM=today_num),
mode="prob", df=NA))
multinom_preds_today_avg
# LINEAGE2 prob SE df asymp.LCL asymp.UCL
# 1 B.1.1.7 7.390771e-02 1.851381e-02 NA 3.762131e-02 1.101941e-01
# 2 B.1.177+ 7.475671e-06 1.346004e-05 NA -1.890553e-05 3.385687e-05
# 3 B.1.351 1.627962e-04 1.493915e-04 NA -1.300057e-04 4.555982e-04
# 4 P.1 2.261638e-03 1.048686e-03 NA 2.062506e-04 4.317025e-03
# 5 B.1.617.1 4.057948e-24 1.263736e-22 NA -2.436298e-22 2.517457e-22
# 6 B.1.617.2 8.943854e-01 2.620214e-02 NA 8.430302e-01 9.457407e-01
# 7 other 2.927494e-02 9.189352e-03 NA 1.126414e-02 4.728574e-02
# % non-B.1.1.7
colSums(multinom_preds_today_avg[-1, c("prob","asymp.LCL","asymp.UCL")])
# prob asymp.LCL asymp.UCL
# 0.9260923 0.8543517 0.9978329
# PLOT MULTINOMIAL FIT
# extrapolate = 30
date.from = as.numeric(as.Date("2021-02-01"))
date.to = as.numeric(as.Date("2021-07-31")) # max(GISAID_sel$DATE_NUM)+extrapolate
# multinomial model predictions (fastest, but no confidence intervals)
predgrid = expand.grid(list(DATE_NUM=seq(date.from, date.to)))
fit_portugal_multi_preds = data.frame(predgrid, as.data.frame(predict(fit2_portugal_multi, newdata=predgrid, type="prob")),check.names=F)
library(tidyr)
library(tidyselect)
fit_portugal_multi_preds = gather(fit_portugal_multi_preds, LINEAGE2, prob, all_of(levels_LINEAGE2), factor_key=TRUE)
fit_portugal_multi_preds$collection_date = as.Date(fit_portugal_multi_preds$DATE_NUM, origin="1970-01-01")
fit_portugal_multi_preds$LINEAGE2 = factor(fit_portugal_multi_preds$LINEAGE2, levels=levels_LINEAGE2)
muller_portugal_mfit = ggplot(data=fit_portugal_multi_preds,
aes(x=collection_date, y=prob, group=LINEAGE2)) +
# facet_wrap(~ STATE) +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE2, group=LINEAGE2), 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")),
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"))),1,1),
limits=as.Date(c("2021-02-01",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 PORTUGAL\n(GISAID data, multinomial fit)")
muller_portugal_mfit
ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_muller plots_multinom fit.png"), width=10, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_muller plots_multinom fit.pdf"), width=10, height=6)
library(ggpubr)
ggarrange(muller_portugal_raw2 + coord_cartesian(xlim=c(as.Date("2021-02-01"),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_portugal_mfit+ggtitle("Multinomial fit"), ncol=1)
ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_muller plots multipanel_multinom fit.png"), width=10, height=10)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_muller plots multipanel_multinom fit.pdf"), width=10, height=10)
# PLOT MODEL FIT WITH DATA & CONFIDENCE INTERVALS
# multinomial model predictions by state with confidence intervals (but slower)
fit_portugal_multi_preds_withCI = data.frame(emmeans(fit2_portugal_multi,
~ LINEAGE2,
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_portugal_multi_preds_withCI$collection_date = as.Date(fit_portugal_multi_preds_withCI$DATE_NUM, origin="1970-01-01")
fit_portugal_multi_preds_withCI$LINEAGE2 = factor(fit_portugal_multi_preds_withCI$LINEAGE2, levels=levels_LINEAGE2)
fit_portugal_multi_preds2 = fit_portugal_multi_preds_withCI
# on logit scale:
ymin = 0.001
ymax = 0.999
fit_portugal_multi_preds2$asymp.LCL[fit_portugal_multi_preds2$asymp.LCL<ymin] = ymin
fit_portugal_multi_preds2$asymp.UCL[fit_portugal_multi_preds2$asymp.UCL<ymin] = ymin
fit_portugal_multi_preds2$asymp.UCL[fit_portugal_multi_preds2$asymp.UCL>ymax] = ymax
fit_portugal_multi_preds2$prob[fit_portugal_multi_preds2$prob<ymin] = ymin
plot_portugal_mfit_logit = qplot(data=fit_portugal_multi_preds2, x=collection_date, y=prob, geom="blank") +
# facet_wrap(~ STATE) +
geom_ribbon(aes(y=prob, ymin=asymp.LCL, ymax=asymp.UCL, colour=NULL,
fill=LINEAGE2
), alpha=I(0.3)) +
geom_line(aes(y=prob,
colour=LINEAGE2
), alpha=I(1)) +
ylab("Share (%)") +
theme_hc() + xlab("") +
ggtitle("SPREAD OF SARS-CoV2 VARIANTS OF CONCERN IN PORTUGAL\n(baseline surveillance GISAID data, multinomial fit)") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01"))),1,1),
limits=as.Date(c("2021-02-01",NA)), expand=c(0,0)) +
scale_y_continuous( trans="logit", breaks=c(10^seq(-5,0),0.5,0.9,0.99,0.999),
labels = c("0.001","0.01","0.1","1","10","100","50","90","99","99.9")) +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
geom_point(data=data_agbyweek2,
aes(x=collection_date, y=prop, size=total,
colour=LINEAGE2
),
alpha=I(1)) +
scale_size_continuous("total number\nsequenced", trans="sqrt",
range=c(0.5, 5), limits=c(1,max(data_agbyweek2$total)), breaks=c(100,1000,10000)) +
# guides(fill=FALSE) +
# guides(colour=FALSE) +
theme(legend.position = "right") +
xlab("Collection date")+
coord_cartesian(xlim=c(as.Date("2021-02-01"),as.Date(date.to, origin="1970-01-01")), ylim=c(0.001, 0.9901), expand=c(0,0))
plot_portugal_mfit_logit
ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_multinom fit_logit scale.png"), width=10, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_multinom fit_logit scale.pdf"), width=10, height=6)
# on response scale:
plot_portugal_mfit = qplot(data=fit_portugal_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=LINEAGE2
), alpha=I(0.3)) +
geom_line(aes(y=100*prob,
colour=LINEAGE2
), alpha=I(1)) +
ylab("Share (%)") +
theme_hc() + xlab("") +
ggtitle("SPREAD OF SARS-CoV2 VARIANTS OF CONCERN IN PORTUGAL\n(baseline surveillance 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")),
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"))),1,1),
limits=as.Date(c("2021-02-01",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("2021-02-01",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_agbyweek2,
aes(x=collection_date, y=100*prop, size=total,
colour=LINEAGE2
),
alpha=I(1)) +
scale_size_continuous("total number\nsequenced", trans="sqrt",
range=c(0.5, 5), limits=c(1,max(data_agbyweek2$total)), breaks=c(100,1000,10000)) +
# guides(fill=FALSE) +
# guides(colour=FALSE) +
theme(legend.position = "right") +
xlab("Collection date")
plot_portugal_mfit
ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_multinom fit_response scale.png"), width=10, height=6)
# ggsave(file=paste0(".\\plots\\",plotdir,"\\portugal_multinom fit_response scale.pdf"), width=10, height=6)
# PLOTS OF NEW CASES PER DAY BY VARIANT & EFFECTIVE REPRODUCTION NUMBER BY VARIANT THROUGH TIME ####
# load case data
library(covidregionaldata)
library(dplyr)
library(ggplot2)
library(scales)
cases_tot = as.data.frame(get_national_data(countries = "Portugal"))
cases_tot = cases_tot[cases_tot$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)
# smooth out weekday effects in case nrs using GAM (if testing data is available one could correct for testing intensity as well)
fit_cases = gam(cases_new ~ s(DATE_NUM, bs="cs", k=20, fx=F) +
WEEKDAY, # +
# BANKHOLIDAY,
# s(TESTS_ALL, bs="cs", k=8, fx=F),
family=poisson(log), data=cases_tot,
)
BIC(fit_cases)
# STACKED AREA CHART OF NEW CASES BY VARIANT (MULTINOMIAL FIT MAPPED ONTO CASE DATA) ####
fit_portugal_multi_preds_withCI$totcases = cases_tot$cases_new[match(round(fit_portugal_multi_preds_withCI$DATE_NUM),cases_tot$DATE_NUM)]
fit_portugal_multi_preds_withCI$cases = fit_portugal_multi_preds_withCI$totcases * fit_portugal_multi_preds_withCI$prob
fit_portugal_multi_preds_withCI$cases[fit_portugal_multi_preds_withCI$cases<=0.001] = NA
cases_emmeans = as.data.frame(emmeans(fit_cases, ~ DATE_NUM, at=list(DATE_NUM=seq(date.from, date.to, by=0.5), BANHOLIDAY="no"), type="response"))
fit_portugal_multi_preds_withCI$smoothed_totcases = cases_emmeans$rate[match(fit_portugal_multi_preds_withCI$DATE_NUM,cases_emmeans$DATE_NUM)]
fit_portugal_multi_preds_withCI$smoothed_cases = fit_portugal_multi_preds_withCI$smoothed_totcases * fit_portugal_multi_preds_withCI$prob
fit_portugal_multi_preds_withCI$smoothed_cases[fit_portugal_multi_preds_withCI$smoothed_cases<=0.001] = NA
ggplot(data=fit_portugal_multi_preds_withCI[fit_portugal_multi_preds_withCI$collection_date>=as.Date("2021-02-18"),],
aes(x=collection_date, y=cases, group=LINEAGE2)) +
# facet_wrap(~ REGION, scale="free", ncol=3) +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE2, group=LINEAGE2), position="stack") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01"))),1,1),
# limits=c(as.Date("2021-03-01"),max(cases_tot$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 PORTUGAL\n(case data & multinomial fit to baseline surveillance GISAID data)") +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
coord_cartesian(xlim=c(as.Date("2021-02-18"),NA))
ggsave(file=paste0(".\\plots\\",plotdir,"\\cases per day_stacked area multinomial fit raw case data.png"), width=8, height=6)
ggplot(data=fit_portugal_multi_preds_withCI[fit_portugal_multi_preds_withCI$collection_date>=as.Date("2021-02-18")&
fit_portugal_multi_preds_withCI$collection_date<=max(cases_tot$date),],
aes(x=collection_date-7, y=smoothed_cases, group=LINEAGE2)) +
# facet_wrap(~ REGION, scale="free", ncol=3) +
geom_area(aes(lwd=I(1.2), colour=NULL, fill=LINEAGE2, group=LINEAGE2), position="stack") +
scale_x_continuous(breaks=as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01")),
labels=substring(months(as.Date(c("2020-01-01","2020-02-01","2020-03-01","2020-04-01","2020-05-01","2020-06-01","2020-07-01","2020-08-01","2020-09-01","2020-10-01","2020-11-01","2020-12-01","2021-01-01","2021-02-01","2021-03-01","2021-04-01","2021-05-01","2021-06-01","2021-07-01"))),1,1),
# limits=c(as.Date("2021-03-01"),today),
expand=c(0,0)) +
# guides(color = guide_legend(reverse=F, nrow=1, byrow=T), fill = guide_legend(reverse=F, nrow=1, byrow=T)) +
theme_hc() + theme(legend.position="right") +
ylab("New confirmed cases per day (smoothed)") + xlab("Date of infection") +
ggtitle("NEW CONFIRMED SARS-CoV2 CASES PER DAY BY VARIANT\nIN PORTUGAL\n(case data & multinomial fit to baseline surveillance GISAID data)") +
scale_fill_manual("variant", values=lineage_cols2) +
scale_colour_manual("variant", values=lineage_cols2) +
coord_cartesian(xlim=c(as.Date("2021-02-18"),max(cases_tot$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")),
labels=c("M","A","M","J","J","A","S","O","N","D","J","F","M","A","M","J","J")) +
# scale_y_continuous(limits=c(1/2, 2), trans="log2") +
geom_hline(yintercept=1, colour=I("red")) +
ggtitle("Re IN PORTUGAL 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_portugal_multi, ~ LINEAGE2 , at=list(DATE_NUM=d), type="response"))$prob # important: these should sum to 1
# wt = rep(1/length(levels_VARIANTS), length(levels_VARIANTS)) # this would give equal weights, equivalent to emmeans:::eff.emmc(levs=levels_LINEAGE2)
cons = lapply(seq_along(wt), function (i) { con = -wt; con[i] = 1 + con[i]; con })
names(cons) = seq_along(cons)
EMT = emtrends(fit2_portugal_multi, ~ LINEAGE2 , by=c("DATE_NUM"),
var="DATE_NUM", mode="latent",
at=list(DATE_NUM=d))
out = as.data.frame(confint(contrast(EMT, cons), adjust="none", df=NA))
# sum(out$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_LINEAGE2),
labels=levels_LINEAGE2)
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$variant, levels=c(levels_LINEAGE2,"avg"))
above_avg_r_variants$prob = fit_portugal_multi_preds_withCI$prob[match(interaction(above_avg_r_variants$DATE_NUM,
above_avg_r_variants$variant),
interaction(fit_portugal_multi_preds_withCI$DATE_NUM,
fit_portugal_multi_preds_withCI$LINEAGE2))]
above_avg_r_variants2 = above_avg_r_variants
ymax = 2
ymin = 1/2
above_avg_r_variants2$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")),
labels=c("M","A","M","J","J","A","S","O","N","D","J","F","M","A","M","J","J")) +
# 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 PORTUGAL\nAT MOMENT OF INFECTION\n(based on case data & multinomial fit to\nbaseline surveillance 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("2021-02-18"),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)