-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathWHALB_model_utilities.R
3689 lines (3271 loc) · 130 KB
/
WHALB_model_utilities.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
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# ----Settings----
#* Environment----
if(exists("seed")){
stanseed = seed
set.seed(seed = seed)
} else {
stanseed = round(rnorm(1,0,100),0)
}
library("tidyverse")
library("rstan")
library("rstanarm")
library("loo")
options(mc.cores = parallel::detectCores())
rstan_options(auto_write = TRUE)
library("shinystan")
later:::ensureInitialized()
options(scipen = 10)
library("readxl")
library("ggpubr")
library("reshape2")
library("EnvStats")
library("extraDistr")
# ----Utility functions------
#* general-------
dnormalize_minmax <- function(d){ # normalize dose to 0 and 1
dmin <- min(d) # applicable to single study and meta regression
dmax <- max(d)
dm <- (d-dmin)/(dmax-dmin)
return(dm)
}
dnormalize_meanSD <- function(d){ # normalize to standardized Z
dmean <- mean(d)
dsd <- sd(d)
dm <- (d-dmean)/dsd
return(dm)
}
get_ysdL <- function(ymean,ysd){ # estimated summarized ysd at log scale
CV <- ysd / ymean
ysdL <- sqrt(log(CV^2 +1)) # based on summarized ymean and ysd
return(ysdL) # at regular scale
}
get_ymeanL <- function(ymean,ysd){ # estimate summarized ymean at log scale
CV <- ysd / ymean
ymeanL <- log(ymean/sqrt(1+CV^2))
return(ymeanL)
}
get_ysd <- function(ymean,ysdL){ # anti of get_ysdL
ysd <- ymean * sqrt(exp(ysdL^2)-1) # applicable to RR data
return(ysd) # get ysd regular from ysd log
}
get_Linear <- function(x,a,b,...){
# calculate response given Linear parameters
# can accomendate raw and normalized dose
# but normalized dose recommended
# parameters must be vector of length 1
if(length(a) != 1) stop("length of a not 1")
if(length(b) != 1) stop("length of b not 1")
x <- unlist(x) # x can be a vector
# returned vector has the same length of x
y <- numeric(length(x))
y <- a + b * x
return(y)
}
get_Power <- function(x,a,b,g,...){
# calculate response given Power parameters
# can accomendate raw and normalized dose
# but normalized dose recommended
# parameters must be real numbers
if(length(a) != 1) stop("length of a not 1")
if(length(b) != 1) stop("length of b not 1")
if(length(g) != 1) stop("length of g not 1")
x <- unlist(x) # x can be a vector
y <- numeric(length = length(x))
y <- a + b * Re(as.complex(x)^g)
return(y)
}
get_MM <- function(x,a,b,c,...){
# calculate response given Power parameters
# can accomendate raw and normalized dose
# but normalized dose recommended
# parameters must be real numbers
if(length(a) != 1) stop("length of a not 1")
if(length(b) != 1) stop("length of b not 1")
if(length(c) != 1) stop("length of c not 1")
x <- unlist(x) # x can be a vector
y <- numeric(length = length(x))
y <- a + b * x / (c + x)
return(y)
}
get_Expo5 <- function(x,a,b,c,d,...){
# calculate response given Power parameters
# can accomendate raw and normalized dose
# but normalized dose recommended
# parameters must be real numbers
if(length(a) != 1) stop("length of a not 1")
if(length(b) != 1) stop("length of b not 1")
if(length(c) != 1) stop("length of c not 1")
if(length(d) != 1) stop("length of d not 1")
x <- unlist(x) # x can be a vector
y <- numeric(length = length(x))
y <- a * (c - (c-1) * exp(-(b*x)^d))
return(y)
}
get_Hill <- function(x,a,b,c,g,...){
# calculate response given Hill parameters
# can accommodate raw and normalized dose
# but normalized dose recommended
# parameters must be real numbers
if(length(a) != 1) stop("length of a not 1")
if(length(b) != 1) stop("length of b not 1")
if(length(c) != 1) stop("length of c not 1")
if(length(g) != 1) stop("length of g not 1")
x <- unlist(x) # x can be a vector
# returned vector has the same length of x
y <- numeric(length(x))
y <- a + (b * Re(as.complex(x)^g))/(c^g + Re(as.complex(x)^g))
return(y)
}
getRRSE <- function(RR.l,RR.u){
SE <- (log(RR.u)-log(RR.l))/2/1.96
SE[is.nan(SE)] <- 0
return(SE)
}
fit_Linear <- function(pars,input,y){
yhat <- pars[1] + pars[2] * input
SSE <- sum((yhat - y)^2)
if(is.finite(SSE)) {return(SSE)} else {
return(1000 + sum(pars^2))
}
}
fit_Hill <- function(pars,input,y){
yhat <- pars[1] + (pars[2] * input^pars[4]) / (pars[3]^pars[4] + input^pars[4])
SSE <- sum((yhat - y)^2)
if(is.finite(SSE)) {return(SSE)} else {
return(1000 + sum(pars^2))
}
}
fit_Power <- function(pars,input,y){
yhat <- pars[1] + pars[2] * (input^pars[3])
SSE <- sum((yhat - y)^2)
if(is.finite(SSE)) {return(SSE)} else {
return(1000 + sum(pars^2))
}
}
fit_MM <- function(pars,input,y){
yhat <- pars[1] + pars[2] * input / (pars[3] + input)
SSE <- sum((yhat - y)^2)
if(is.finite(SSE)) {return(SSE)} else {
return(1000 + sum(pars^2))
}
}
fit_Expo5 <- function(pars,input,y){
yhat <- pars[1] * (pars[3] - (pars[3]-1) * exp(-1 * (pars[2] * input)^pars[4]))
SSE <- sum((yhat - y)^2)
if(is.finite(SSE)) {return(SSE)} else {
return(1000 + sum(pars^2))
}
}
# *Data Treatment-------
# Simulate meta DR data based on four DR models
fn_metaDRsimulation <- function(S = 10 , G = 3,Sub = c(250,210,180),
Dose = c(10,50,90),
ybound = c(exp(-10),20),
sigmavalue = 0.2){
N <- S*G
Index <- rep(1:S,each = G)
Study <- LETTERS[Index]
# alldose <- replicate(S,seq(from = 0, to = max(dose), lengt = 1E3))
Nsub <- ceiling(replicate(S,Sub+runif(G,0,60)))
dose <- replicate(S,Dose + runif(G,-1,1)*10)
x_bound <- c(min(dose),max(dose))
y_bound <- sort(runif(2,min = ybound[1],max = ybound[2]))
noisesize <- 0.1
# Linear
popfit_linear <- optim(c(1,1),fit_Linear, input = x_bound, y = y_bound,
method = c("L-BFGS-B"),lower = c(0,0))
pars_linear <- popfit_linear$par
# get_Linear(x_bound,pars_linear[1],pars_linear[2])
a_Linear <- pars_linear[1] + rnorm(S,0,noisesize * pars_linear[1])
b_Linear <- pars_linear[2] + rnorm(S,0,noisesize * pars_linear[2])
# Hill
popfit_Hill <- optim(c(1,1,50,1),fit_Hill, input = x_bound, y = y_bound,
method = c("L-BFGS-B"),lower = c(0,0,0,0),upper = c(Inf,Inf,Inf,18))
pars_Hill <- popfit_Hill$par
# get_Hill(x_bound,pars_Hill[1],pars_Hill[2],pars_Hill[3],pars_Hill[4])
a_Hill <- pars_Hill[1] + rnorm(S,0,noisesize * pars_Hill[1])
b_Hill <- pars_Hill[2] + rnorm(S,0,noisesize * pars_Hill[2])
c_Hill <- pars_Hill[3] + rnorm(S,0,noisesize * pars_Hill[3])
g_Hill <- pars_Hill[4] + rnorm(S,0,noisesize * pars_Hill[4])
# Power
popfit_Pow <- optim(c(1,1,1),fit_Power, input = x_bound, y = y_bound,
method = "L-BFGS-B",lower = c(0,0,0), upper = c(Inf,Inf,18))
pars_Pow<- popfit_Pow$par
# get_Power(x_bound,pars_Pow[1],pars_Pow[2],pars_Pow[3])
a_Pow <- pars_Pow[1] + rnorm(S,0,noisesize * pars_Pow[1])
b_Pow <- pars_Pow[2] + rnorm(S,0,noisesize * pars_Pow[2])
g_Pow <- pars_Pow[3] + rnorm(S,0,noisesize * pars_Pow[3])
# MM
popfit_MM <- optim(c(0.5,0.5,0.5),fit_MM, input = x_bound, y = ybound,
method = "L-BFGS-B",lower = c(0,0,0)
)
pars_MM <- popfit_MM$par
a_MM <- pars_MM[1] + rnorm(S,0,noisesize * pars_MM[1])
b_MM <- pars_MM[2] + rnorm(S,0,noisesize * pars_MM[2])
c_MM <- pars_MM[3] + rnorm(S,0,noisesize * pars_MM[3])
# Exponential 5
popfit_Expo5 <- optim(c(0.5,0.5,1,1),fit_Expo5, input = x_bound, y = y_bound,
method = "L-BFGS-B",lower = c(0,0,1,1), upper = c(Inf,Inf,Inf,18))
pars_Expo5 <- popfit_Expo5$par
# get_Expo5(x_bound,pars_Expo5[1],pars_Expo5[2],pars_Expo5[3],pars_Expo5[4])
a_Expo5 <- pars_Expo5[1] + rnorm(S,0,noisesize * pars_Expo5[1])
b_Expo5 <- pars_Expo5[2] + rnorm(S,0,noisesize * pars_Expo5[2])
c_Expo5 <- pars_Expo5[3] + rnorm(S,0,noisesize * pars_Expo5[3])
d_Expo5 <- pars_Expo5[4] + rnorm(S,0,noisesize * pars_Expo5[4])
# Plotting and simulation fd
fd_Linear <- vector("list",S)
fd_Hill <- vector("list",S)
fd_MM <- vector("list",S)
fd_Power <- vector("list",S)
fd_Expo5 <- vector("list",S)
xdose <- dose # use dose for simulation/ alldose for plotting
# plot(0,xlim = c(0,max(dose)), ylim = c(0,y_bound[2]))
# for(s in 1:S){
# fd_Linear[[s]] <- get_Linear(xdose[,s],a = a_Linear[s], b = b_Linear[s])
# lines(xdose[,s],fd_Linear[[s]],col = "blue")
# fd_Hill[[s]] <- get_Hill(xdose[,s],a = a_Hill[s], b = b_Hill[s], c = c_Hill[s], g = g_Hill[s])
# lines(xdose[,s],fd_Hill[[s]], col = "red")
# fd_Power[[s]] <- get_Power(xdose[,s], a = a_Pow[s], b = b_Pow[s], g = g_Pow[s])
# lines(xdose[,s],fd_Power[[s]], col = "green")
# fd_Expo5[[s]] <- get_Expo5(xdose[,s], a = a_Expo5[s], b = b_Expo5[s], c = c_Expo5[s], d = d_Expo5[s])
# lines(xdose[,s],fd_Expo5[[s]],col = "yellow")
# }
for(s in 1:S){
fd_Linear[[s]] <- get_Linear(xdose[,s],a = a_Linear[s], b = b_Linear[s])
fd_Hill[[s]] <- get_Hill(xdose[,s],a = a_Hill[s], b = b_Hill[s], c = c_Hill[s], g = g_Hill[s])
fd_Power[[s]] <- get_Power(xdose[,s], a = a_Pow[s], b = b_Pow[s], g = g_Pow[s])
fd_MM[[s]] <- get_MM(xdose[,s], a = a_MM[s], b = b_MM[s], c = c_MM[s])
fd_Expo5[[s]] <- get_Expo5(xdose[,s], a = a_Expo5[s], b = b_Expo5[s], c = c_Expo5[s], d = d_Expo5[s])
}
y_Linear <- vector("list",N)
y_Hill <- vector("list",N)
y_MM <- vector("list",N)
y_Power <- vector("list",N)
y_Expo5 <- vector("list",N)
for(g in 1:N){
y_Linear[[g]] <- rlnorm(Nsub[g],meanlog = log(fd_Linear[[Index[g]]][g-3*(Index[g]-1)]),sdlog = sigmavalue)
y_Hill[[g]] <- rlnorm(Nsub[g],meanlog = log(fd_Hill[[Index[g]]][g-3*(Index[g]-1)]), sdlog = sigmavalue)
y_Power[[g]] <- rlnorm(Nsub[g], meanlog = log(fd_Power[[Index[g]]][g-3*(Index[g]-1)]), sdlog = sigmavalue)
y_Expo5[[g]] <- rlnorm(Nsub[g], meanlog = log(fd_Expo5[[Index[g]]][g-3*(Index[g]-1)]), sdlog = sigmavalue)
y_MM[[g]] <- rlnorm(Nsub[g], meanlog = log(fd_MM[[Index[g]]][g-3*(Index[g]-1)]), sdlog = sigmavalue)
}
y_Linear_summary <- lapply(y_Linear,function(x){
cbind(mean(x),sd(x))
})
y_Hill_summary <- lapply(y_Hill,function(x){
cbind(mean(x),sd(x))
})
y_Power_Summary <- lapply(y_Power,function(x){
cbind(mean(x),sd(x))
})
y_MM_Summary <- lapply(y_MM,function(x){
cbind(mean(x),sd(x))
})
y_Expo5_Summary <- lapply(y_Expo5, function(x){
cbind(mean(x),sd(x))
})
df_simu_meta_Linear <- data.frame(
Index = Index,
Study = Study,
dose = as.vector(dose),
Nsub = as.vector(Nsub),
ymean = unlist(lapply(y_Linear_summary,function(x) x[,1])),
ysd = unlist(lapply(y_Linear_summary,function(x) x[,2]))
)
df_simu_meta_Hill <- data.frame(
Index = Index,
Study = Study,
dose = as.vector(dose),
Nsub = as.vector(Nsub),
ymean = unlist(lapply(y_Hill_summary,function(x) x[,1])),
ysd = unlist(lapply(y_Hill_summary,function(x) x[,2]))
)
df_simu_meta_Power <- data.frame(
Index = Index,
Study = Study,
dose = as.vector(dose),
Nsub = as.vector(Nsub),
ymean = unlist(lapply(y_Power_Summary,function(x) x[,1])),
ysd = unlist(lapply(y_Power_Summary,function(x) x[,2]))
)
df_simu_meta_MM <- data.frame(
Index = Index,
Study = Study,
dose = as.vector(dose),
Nsub = as.vector(Nsub),
ymean = unlist(lapply(y_MM_Summary,function(x) x[,1])),
ysd = unlist(lapply(y_MM_Summary,function(x) x[,2]))
)
df_simu_meta_Expo5 <- data.frame(
Index = Index,
Study = Study,
dose = as.vector(dose),
Nsub = as.vector(Nsub),
ymean = unlist(lapply(y_Expo5_Summary,function(x) x[,1])),
ysd = unlist(lapply(y_Expo5_Summary,function(x) x[,2]))
)
baseinfo <- list(
x_bound,y_bound,
a_Linear,b_Linear,
a_Pow,b_Pow,g_Pow,
a_MM,b_MM,c_MM,
a_Hill,b_Hill,c_Hill,g_Hill,
a_Expo5,b_Expo5,c_Expo5,d_Expo5
)
ls_data <- list(
Linear = df_simu_meta_Linear,
Power = df_simu_meta_Power,
MM = df_simu_meta_MM,
Hill = df_simu_meta_Hill,
Expo5 = df_simu_meta_Expo5,
baseinfo = baseinfo
)
return(ls_data)
}
# Extract Simulated Data by Model
fn_getsimuData <- function(ls_data,DSName){
df_input <- as.data.frame(ls_data[DSName])
colnames(df_input) <- stringr::str_replace(colnames(df_input),pattern = paste0(DSName,"."),replacement = "")
return(df_input)
}
# reorder data and output for testing
ReorderData <- function(df_input,OrderNew){
# extract data by group into a list, which is ordered by the new order
# then concatenate into a dataframe
# number of groups
S <- length(OrderNew)
ls_all <- list() # list for storage
IndexNew <- integer()
# extract data
for(s in 1:S){
ls_all[[s]] <- df_input %>% filter(Study == OrderNew[s])
# store the new index
IndexNew <- c(IndexNew,rep(s, nrow(ls_all[[s]])))
}
# concatenate
df_output <- do.call("rbind",ls_all)
# Re-indexing
df_output <- df_output %>% mutate(
Index = IndexNew
)
return(df_output)
}
DFtoMatrix <- function(df_input){ # convert long data frame to wide matrices
# add dose group index
df_temp <- df_input %>% group_by(Study) %>%
mutate(GID = row_number())
# df_temp <- filter(df_temp,GID != 1)
G <- as.integer(table(df_temp$Study)) # # of dose groups each study
Gmax <- max(G) # largest number of dose group
S <- max(df_temp$Index) # number of study
# Ginits <- integer(length = S)
# for(s in 1:S){
# Ginits[s] = ifelse(s == 1, 1, sum(G[1:(s-1)])+1)
# }
# matrices to store dose, mean SD and N
# one study per row
# one dose group per column
mtx_doseD <- matrix(NA,nrow = S,ncol = Gmax)
mtx_dose <- mtx_doseD
mtx_ymeanL <- mtx_doseD
mtx_ysdL <- mtx_doseD
mtx_n <- mtx_doseD
for(s in 1:S){
# extract data by study
df_study <- df_temp %>% filter(Index == s)
g_temp <- nrow(df_study) # number of dose groups in the study
# extract dose, ymean ysd and N
vec_doseD_raw <- df_study$dose_minmax # vector of standarized dose
vec_dose_raw <- df_study$dose # dose
vec_ymeanL_Raw <- df_study$RRlog # vector of mean at log
vec_ysdL_raw <- df_study$ysdL # vector of SD at log
vec_n_raw <- df_study$N # vector of N
# supplement with NA if # dose group less than max # dose group
if(g_temp < Gmax){
NACount <- Gmax - g_temp
vec_doseD <- c(vec_doseD_raw,rep(2,NACount))
vec_dose <- c(vec_dose_raw,rep(2,NACount))
vec_ymeanL <- c(vec_ymeanL_Raw,rep(2,NACount))
vec_ysdL <- c(vec_ysdL_raw,rep(2,NACount))
vec_n <- c(vec_n_raw,rep(2,NACount))
} else {
vec_doseD <- vec_doseD_raw
vec_dose <- vec_dose_raw
vec_ymeanL <- vec_ymeanL_Raw
vec_ysdL <- vec_ysdL_raw
vec_n <- vec_n_raw
}
# fill in completed data into matrix
mtx_doseD[s,] <- vec_doseD
mtx_dose[s,] <- vec_dose
mtx_ymeanL[s,] <- vec_ymeanL
mtx_ysdL[s,] <- vec_ysdL
mtx_n[s,] <- vec_n
}
# Output in a list
ls_output <- list(mtx_n,mtx_doseD,mtx_dose,mtx_ymeanL,mtx_ysdL,G,S,
# Ginits,
Gmax)
names(ls_output) <- c("mtx_N","mtx_doseD","mtx_dose","mtx_ymeanL","mtx_ysdL","G","S",
# "Ginits",
"Gmax")
return(ls_output)
}
PrepareRRData <- function(df_input){
Index <- df_input$Index
Study <- df_input$Study
dose <- df_input$dose
Nsub <- df_input$Nsub
ymean <- df_input$RR
ymeanL <- log(ymean)
ysdL <- getRRSE(RR.l = df_input$RR.l,RR.u = df_input$RR.u)
ysd <- get_ysd(ymean,ysdL)
doseZ <- dnormalize_minmax(dose)
N <- nrow(df_input)
S <- length(unique(df_input$Index))
Gcount <- df_input %>% group_by(Index) %>% count()
G <- as.data.frame(ungroup(Gcount))[,2]
Data <- list(N = N,S = S,G = G,
Index = Index,Study = Study,Nsub = Nsub,
ymean = ymean,ymeanL = ymeanL,
ysd = ysd, ysdL = ysdL,
doseZ = doseZ,dose = dose)
return(Data)
}
PrepareNRRLnormData <- function(df_input){
Index <- df_input$Index
Study <- df_input$Study
dose <- df_input$dose
Nsub <- df_input$Nsub
ymean <- df_input$ymean
ysd <- df_input$ysd
ymeanL <- get_ymeanL(ymean,ysd)
ysdL <- get_ysdL(ymean,ysd)
doseZ <- dnormalize_minmax(dose)
N <- nrow(df_input)
S <- length(unique(df_input$Index))
Gcount <- df_input %>% group_by(Index) %>% count()
G <- as.data.frame(ungroup(Gcount))[,2]
Data <- list(N = N,S = S,G = G,
Index = Index,Study = Study,Nsub = Nsub,
ymean = ymean,ymeanL = ymeanL,
ysd = ysd, ysdL = ysdL,
doseZ = doseZ,dose = dose
)
return(Data)
}
PrepareLnormData <- function(df_input, RR=F){
if(RR){
PreparedData <- PrepareRRData(df_input)
} else {
PreparedData <- PrepareNRRLnormData(df_input)
}
return(PreparedData)
}
censored_lnorm_up_close <- function(low,up,Nsub){
G <- length(low)
if(is.na(up[G])) warning("The upper bound of the last group is infinity")
q <- Nsub / sum(Nsub)
fit_lnorm_up_close <- function(par,low,up,Nsub){
# LS <- vector(length = G)
# for(i in 1:G){
# LS[i] <- (plnorm(log(up[i]),par[1],par[2]) - plnorm(log(low[i]),par[1],par[2]) - q[i])^2
# }
# SLS <- sum(LS)
# return(SLS)
LL <- vector(length = G)
for(i in 1:G){
LL[i] <- q[i] * log(plnorm(up[i],par[1],par[2])-plnorm(low[i],par[1],par[2]))
}
MLE <- -sum(LL)
}
fit <- optim(
# par = c(log(0.1),5),
par = c(0,1),
fn = fit_lnorm_up_close,
low = low, up = up , Nsub = Nsub
)
pars_lnorm <- fit$par
return(pars_lnorm)
}
censored_norm_up_close <- function(low,up,Nsub){
G <- length(low)
if(is.na(up[G])) warning("The upper bound of the last group is infinity")
q <- Nsub / sum(Nsub)
fit_norm_up_close <- function(par,low,up,Nsub){
# LS <- vector(length = G)
# for(i in 1:G){
# LS[i] <- (plnorm(log(up[i]),par[1],par[2]) - plnorm(log(low[i]),par[1],par[2]) - q[i])^2
# }
# SLS <- sum(LS)
# return(SLS)
LL <- vector(length = G)
for(i in 1:G){
LL[i] <- q[i] * log(pnorm(up[i],par[1],par[2])-pnorm(low[i],par[1],par[2]))
}
MLE <- -sum(LL)
}
fit <- optim(
# par = c(log(0.1),5),
par = c(0,1),
fn = fit_norm_up_close,
low = low, up = up , Nsub = Nsub
)
pars_norm <- fit$par
return(pars_norm)
}
censored_lnorm_up_open <- function(low,up,Nsub){
G <- length(low)
q <- Nsub / sum(Nsub)
if(!is.na(up[G])) warning("The upper bound of the last group is infinity")
fit_lnorm_up_open <- function(par,low,up,Nsub){
# LS <- vector(length = G)
# for(i in 1:(G-1)){
# LS[i] <- (plnorm(log(up[i]),par[1],par[2]) - plnorm(log(low[i]),par[1],par[2]) - q[i])^2
# }
# LS[G] <- (1-plnorm(log(low[G]),par[1],par[2]) - q[G])^2
# SLS <- sum(LS)
# return(SLS)
LL <- vector(length = G)
for(i in 1:(G-1)){
LL[i] <- q[i] * log(plnorm(up[i],par[1],par[2])-plnorm(low[i],par[1],par[2]))
}
LL[G] <- q[G] * log(1-plnorm(low[G],par[1],par[2]))
MLE <- -sum(LL)
}
fit <- optim(
# par = c(log(0.1),5),
par = c(0,1),
fn = fit_lnorm_up_open,
low = low, up = up , Nsub = Nsub
)
pars_lnorm <- fit$par
return(pars_lnorm)
}
censored_norm_up_open <- function(low,up,Nsub){
G <- length(low)
q <- Nsub / sum(Nsub)
if(!is.na(up[G])) warning("The upper bound of the last group is infinity")
fit_norm_up_open <- function(par,low,up,Nsub){
# LS <- vector(length = G)
# for(i in 1:(G-1)){
# LS[i] <- (plnorm(log(up[i]),par[1],par[2]) - plnorm(log(low[i]),par[1],par[2]) - q[i])^2
# }
# LS[G] <- (1-plnorm(log(low[G]),par[1],par[2]) - q[G])^2
# SLS <- sum(LS)
# return(SLS)
LL <- vector(length = G)
for(i in 1:(G-1)){
LL[i] <- q[i] * log(pnorm(up[i],par[1],par[2])-pnorm(low[i],par[1],par[2]))
}
LL[G] <- q[G] * log(1-pnorm(low[G],par[1],par[2]))
MLE <- -sum(LL)
}
fit <- optim(
# par = c(log(0.1),5),
par = c(0,1),
fn = fit_norm_up_open,
low = low, up = up , Nsub = Nsub
)
pars_norm <- fit$par
return(pars_norm)
}
#* Hill specific------------
get_prior_a <- function(ymean,ysd){ # get hyperparameters for intercept a
ymax <- max(ymean) # applicable to normalized response
idmax <- which(ymean == max(ymean))
ysd_ymax <- ysd[idmax]
a_upper <- 2 * (ymax + 2 * ysd_ymax)
prior_a <- c(0,a_upper)
return(prior_a)
}
get_prior_b <- function(ymean,ysd,dose){ # get hyperparameter for slope b
idmax <- which(dose == max(dose)) # applicable to normalized response
ymean_maxd <- ymean[idmax]
ysd_maxd <- ysd[idmax]
idmin <- which(dose == min(dose))
ymean_mind <- ymean[idmin]
ysd_mind <- ysd[idmin]
b_slope <- (ymean_maxd+2*ysd_maxd-ymean_mind-2*ysd_mind)/
(dose[idmax]-dose[idmin])
prior_b <- c(0,b_slope * 5)
return(prior_b)
}
#* meta Hill-----------
# get hyperparameter intercept a in meta data
get_prior_meta_a <- function(index,ymean,ysd){
# S is the number of studies in the meta data
S <- length(unique(index))
# estimate hyperparameter a for each study
meta_a <- vector(mode = "numeric",length = S)
for(i in 1:S){
# id_temp is a vector of logical indicator of selected study
id_temp <- which(index == i)
meta_a[i] <- get_prior_a(ymean[id_temp],ysd[id_temp])[2]
}
return(meta_a)
}
# get hyperparameter slope b in meta data
get_prior_meta_b <- function(index,ymean,ysd,dose){
# estimate b for each study
# S is the number of study in the meta data
S <- length(unique(index))
meta_b <- vector(mode = "numeric",length = S)
for(i in 1:S){
id_temp <- which(index == i)
meta_b[i] <- get_prior_b(ymean[id_temp],ysd[id_temp],dose[id_temp])[2]
}
return(meta_b)
}
# Model fitting---------
get_metaStanfit <- function(modelname,dataName,DataList,
inits = NULL){
iterEff <- (samplingiter*(1-warmup)*Nchain)
# Priors
# Calculate data-based control values for hyperparameters
ybound <- c(
max(DataList$ymean),
sd(DataList$ymean)
)
a_bound <- log(ybound[1] + 3 * ybound[2])*2
Xrange <- sort(c(max(DataList$doseZ),min(DataList$doseZ)))
xbound <- c(
Xrange[1],
Xrange[2]
)
source(paste0(fileDir,"BHBMD_priors.R"),local = T)
inits <- ifelse(is.null(inits),"random",inits)
# Compile stanmodels
# assign(paste0("model_",modelname),
# # stan_model(file = paste0("stanmodel_",modelname,".stan"))
# stan_model(model_code = eval(parse(text = paste0("modelstring_",modelname))))
# )
assign(paste0("model_",modelname),
stan_model(file = paste0(fileDir, modelname,".stan")))
SaveFileName <- paste0("stanfit_",modelname,"_",dataName)
assign(SaveFileName,
rstan::sampling(
object = eval(parse(text = paste0("model_",modelname))),
data = append(DataList,
eval(parse(text = paste0("prior_",
modelname)))
),
iter = samplingiter,
chains = Nchain,
warmup = warmup * samplingiter,
thin = thin,
seed = stanseed,
init = inits,
# init_r = 2,
control = list(
# adapt_delta = 0.99,
# max_treedepth = 12
),
cores = parallel::detectCores()
))
# save(list = SaveFileName,file = paste0(SaveFileName,".rdata"))
saveRDS(
object = eval(parse(text = paste0(SaveFileName))),
file = paste0(SaveFileName,".rds")
)
return(eval(parse(text = SaveFileName)))
}
# wrapper to fit all four dose response models given a DataList
fit_allmodels <- function(DSName,DataList){
dataName <- paste0("simu_",DSName)
modelname <- "Lognormal_Summary_Linear_meta_3"
assign(paste0("stanfit_",modelname,"_",dataName),
get_metaStanfit(modelname = modelname,dataName = dataName,DataList = DataList))
modelname <- "Lognormal_Summary_Power_meta_4"
assign(paste0("stanfit_",modelname,"_",dataName),
get_metaStanfit(modelname = modelname,dataName = dataName,DataList = DataList))
modelname <- "Lognormal_Summary_Hill_meta_5"
assign(paste0("stanfit_",modelname,"_",dataName),
get_metaStanfit(modelname = modelname,dataName = dataName,DataList = DataList))
modelname <- "Lognormal_Summary_Expo5_meta_5"
assign(paste0("stanfit_",modelname,"_",dataName),
get_metaStanfit(modelname = modelname,dataName = dataName,DataList = DataList))
assign(
paste0("fitList_",dataName),
list(
Linear = eval(parse(text = paste0("stanfit_Lognormal_Summary_Linear_meta_3_",dataName))),
Hill = eval(parse(text = paste0("stanfit_Lognormal_Summary_Hill_meta_5_",dataName))),
Power = eval(parse(text = paste0("stanfit_Lognormal_Summary_Power_meta_4_",dataName))),
Expo5 = eval(parse(text = paste0("stanfit_Lognormal_Summary_Expo5_meta_5_",dataName)))
)
)
return(eval(parse(text = paste0("fitList_",dataName))))
}
# wrapper to fit four dose response models to a list of simulated data
fit_simuData <- function(simuData,DSName){
df_input <- fn_getsimuData(simuData,DSName)
DataList <- PrepareLnormData(df_input,RR=F)
dataName <- paste0("simu_",DSName)
fitList <- fit_allmodels(DSName = DSName,DataList = DataList)
return(fitList)
}
get_BHBMD_posterior <- function(
stanfit,level,
model = "all", Index = "all"){
if(level == "overarching"){
output <- get_overarching(stanfit,model)
} else if(level == "specific"){
output <- get_specific(stanfit,model,Index)
}
return(output)
}
get_overarching <- function(stanfit,model){
# one model or all four
if(model == "all"){
output <- list(
Linear = get_Linear_overarching(stanfit = stanfit$Linear),
Power = get_Power_overarching(stanfit = stanfit$Power),
Hill = get_Hill_overarching(stanfit = stanfit$Hill),
Expo5 = get_Expo5_overarching(stanfit = stanfit$Expo5)
)
} else {
stanfit <- stanfit[[model]]
output <- get(paste0("get_",model,"_overarching"))(stanfit = stanfit)
}
return(output)
}
get_specific <- function(stanfit,model,Index){
# one model or all four
if(model == "all"){
output <- list(
Linear = get_Linear_specific(stanfit$Linear,Index = Index),
Power = get_Power_specific(stanfit$Power,Index = Index),
Hill = get_Hill_specific(stanfit$Hill,Index = Index),
Expo5 = get_Expo5_specific(stanfit$Expo5,Index = Index)
)
} else {
stanfit <- stanfit[[model]]
output <- get(paste0("get_",model,"_specific"))(stanfit,Index = Index)
}
return(output)
}
get_Linear_overarching <- function(stanfit){
df_posteriors <- as.data.frame(stanfit)
iterEff <- nrow(df_posteriors)
a <- rlnorm(iterEff,df_posteriors$mu_a,df_posteriors$sigma_a)
b <- extraDistr::rhcauchy(iterEff,df_posteriors$scale_b)
df_pars <- data.frame(
a = a,
b = b
)
return(df_pars)
}
get_Power_overarching <- function(stanfit){
df_posteriors <- as.data.frame(stanfit)
iterEff <- nrow(df_posteriors)
a_bound <- 0
source(
file = paste0(fileDir,"BHBMD_priors.R"),local = T
)
a <- rlnorm(iterEff,df_posteriors$mu_a,df_posteriors$sigma_a)
b <- extraDistr::rhcauchy(iterEff,df_posteriors$scale_b)
# g <- rlnorm(iterEff,df_posteriors$mu_g,df_posteriors$sigma_g)
g <- rlnormTrunc(iterEff,df_posteriors$mu_g,df_posteriors$sigma_g,
min = prior_Lognormal_Summary_Power_meta_4$g_constr[1],
max = prior_Lognormal_Summary_Power_meta_4$g_constr[2])
df_pars <- data.frame(
a = a,
b = b,
g = g
)
return(df_pars)
}
get_MM_overarching <- function(stanfit){
df_posteriors <- as.data.frame(stanfit)
iterEff <- nrow(df_posteriors)
a <- rlnorm(iterEff,df_posteriors$mu_a,df_posteriors$sigma_a)
b <- extraDistr::rhcauchy(iterEff,df_posteriors$scale_b)
c <- rlnorm(iterEff,df_posteriors$mu_c,df_posteriors$sigma_c)
df_pars <- data.frame(
a = a,
b = b,
c = c
)
return(df_pars)
}
get_Hill_overarching <- function(stanfit){
df_posteriors <- as.data.frame(stanfit)
iterEff <- nrow(df_posteriors)
a_bound <- 0
source(
file = paste0(fileDir,"BHBMD_priors.R"),local = T
)
a <- rlnorm(iterEff,df_posteriors$mu_a,df_posteriors$sigma_a)
b <- extraDistr::rhcauchy(iterEff,df_posteriors$scale_b)
c <- rlnorm(iterEff,df_posteriors$mu_c,df_posteriors$sigma_c)
g <- rlnormTrunc(iterEff,df_posteriors$mu_g,df_posteriors$sigma_g,
min = prior_Lognormal_Summary_Hill_meta_5$g_constr[1],
max = prior_Lognormal_Summary_Hill_meta_5$g_constr[2])
df_pars <- data.frame(
a = a,
b = b,
c = c,
g = g
)
return(df_pars)
}
get_Expo5_overarching <- function(stanfit){
df_posteriors <- as.data.frame(stanfit)
iterEff <- nrow(df_posteriors)
a <- rlnorm(iterEff,df_posteriors$mu_a,df_posteriors$sigma_a)
b <- extraDistr::rhcauchy(iterEff,df_posteriors$scale_b)
d <- rlnormTrunc(iterEff,df_posteriors$mu_d,df_posteriors$sigma_d,
max = 18)
c <- vector("numeric",iterEff)
for(i in 1:iterEff){
c[i] <- extraDistr::rhcauchy(1,df_posteriors$scale_c)
while (c[i] <= 1) {
c[i] <- extraDistr::rhcauchy(1,df_posteriors$scale_c)
}
}
df_pars <- data.frame(
a = a,
b = b,
c = c,
d = d
)
return(df_pars)
}
get_Linear_specific <- function(stanfit,Index){
df_posteriors <- as.data.frame(stanfit)
iterEff <- nrow(df_posteriors)
## check if index is within range
NIndex <- length(Index)
NStudy <- sum(!is.na(stringr::str_match(colnames(df_posteriors),"a.\\d.")))
if(is.numeric(Index) & max(Index) > NStudy){
stop("Specified study index not within range")
}
if(NIndex == 1 & Index[1] == "all") {NIndex <- NStudy}
## get output
if(is.numeric(Index) & NIndex == 1){
## output one study
output <- data.frame(
a = df_posteriors[,paste0("a[",Index,"]")],
b = df_posteriors[,paste0("b[",Index,"]")]
)
} else if (is.numeric(Index) & NIndex > 1){
## output multiple studies
output <- vector("list",NIndex)
names(output) <- paste0("study",Index)
for(i in 1:NIndex){
output[[i]] <- data.frame(
a = df_posteriors[,paste0("a[",i,"]")],
b = df_posteriors[,paste0("b[",i,"]")]
)
}
} else {
## all studies
output <- vector("list",NIndex)
names(output) <- paste0("study",1:NStudy)
for(i in 1:NIndex){