/
examles_and_figures_1.R
262 lines (190 loc) · 11.3 KB
/
examles_and_figures_1.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
library(ggplot2)
library(gridExtra)
library(DepInfeR)
library(missForest)
library(tidyverse)
library(glmnet)
library(BiocParallel)
library(matrixStats)
library(randomForestSRC)
source('my_helper_funcs.R')
library(ggtext)
# reproducing the figs in main body
mean_fps <- function(protein, X, GP_est){
protein_id = which(colnames(X)==protein)
protein_mean = GP_est$curve_mean[,,protein_id,]
dtfrm <- apply(X=protein_mean, MARGIN = c(2,3), FUN = mean, na.rm=TRUE)
dtfrm <- as.data.frame(cbind(as.vector(t(dtfrm)), rep(seq(0, 1, length.out=10), times=dim(dtfrm)[1])))
colnames(dtfrm) <- c('Contribution', 'Affinity')
dtfrm$Sample <- rep(sapply(1:(dim(dtfrm)[1]/10),
FUN=function(x){paste0('cell', x)}), each=10)
dtfrm$Kinase <- protein
return(dtfrm)
# plot(seq(0, 1, length.out=10), colMeans(protein_mean[,1,], na.rm=TRUE), type='l', ylim=ylim,
# ylab='Function value', xlab = 'Affinity score')
# title(paste0('beatAML, f_ps with protein ', protein, ' of all sample cells'))
# for (i in 2:dim(Y_new)[2]){
# lines(seq(0, 1, length.out=10), colMeans(protein_mean[,i,], na.rm=TRUE), type='l')
# }
}
# beat_AML
# reproducing the figs in main body
load('beatAML_dataset.RData')
load('beatAML_CV.RData')
beatAML_X_original <- tarMat_BeatAML
beatAML_X_new <- tarMat_BeatAML
beatAML_X_nan <- beatAML_X_new[1,1]
beatAML_X_new[beatAML_X_new == beatAML_X_nan] <- NaN
beatAML_Y_new <- viabMat_BeatAML_raw_log
beatAML_Y_original <- viabMat_BeatAML
# run the spike-and-slab GP model, save the results
if (!file.exists('beatAML_GP0_inference.RData')){
# visualizing GP curves
set.seed(314159)
GP_0_z <- my_gibbs_sampler_gp_0(N=120, burn_in=20, thin=1, X=beatAML_X_new, Y=beatAML_Y_new,
inclusion_prob=0.1,
nu=CV_gp_0_log$c0[[which.min(CV_gp_0_log$mse)]],
update_Z=TRUE)
GP_0_para <- GP_para_0(burn_in=20, X=beatAML_X_new, Y=beatAML_Y_new, Z=GP_0_z$Z,
sigma2=GP_0_z$sigma2, gamma2=GP_0_z$gamma2,
nu=CV_gp_0_log$c0[[which.min(CV_gp_0_log$mse)]])
save(GP_0_z, GP_0_para, file = 'beatAML_GP0_inference.RData')
} else{
load('beatAML_GP0_inference.RData')
}
beatAML_CDK6 <- mean_fps('CDK6', beatAML_X_new, GP_0_para)
beatAML_PDXK <- mean_fps('PDXK', beatAML_X_new, GP_0_para)
beatAML_AKT1 <- mean_fps('AKT1', beatAML_X_new, GP_0_para)
beatAML_CDK17 <- mean_fps('CDK17', beatAML_X_new, GP_0_para)
set.seed(314159)
if (!file.exists('DepInfeR_beatAML_res.RData')){
beatAML_DepInfeR_yhat <- MultiLasso(TargetMatrix=beatAML_X_original, ResponseMatrix=beatAML_Y_original, repeats = 100, BPPARAM = bpparam())
} else {load('DepInfeR_beatAML_res.RData')}
beatAML_DepInfeR_Y_hat <- t(t(beatAML_X_original %*% beatAML_DepInfeR_yhat$coefMat) + beatAML_DepInfeR_yhat$intercept)
beatAML_DepInfeR_res <- beatAML_Y_original - t(t(beatAML_X_original %*% beatAML_DepInfeR_yhat$coefMat) + beatAML_DepInfeR_yhat$intercept)
beatAML_GP0_res <- as.vector(beatAML_Y_new - apply(GP_0_para$y_hat, c(1,2), mean))
beatAML_GP0_Y_hat <- as.vector(apply(GP_0_para$y_hat, c(1,2), mean))
beatAML_obs_new <- as.vector(beatAML_Y_new)
beatAML_obs_old <- as.vector(beatAML_Y_original)
beatAML_fit <- data.frame(cbind(as.vector(beatAML_DepInfeR_Y_hat),
as.vector(beatAML_DepInfeR_res),
beatAML_GP0_res, beatAML_GP0_Y_hat, beatAML_obs_new, beatAML_obs_old))
colnames(beatAML_fit) <- c('DepInfeR_yhat', 'DepInfeR_res', 'GP0_res', 'GP0_yhat', 'Yobs_new', 'Yobs_old')
beatAML_fit_fig <- list(ggplot(data = beatAML_fit) + geom_point(aes(x=Yobs_old, y=DepInfeR_yhat)) +
geom_abline(intercept=0, slope=1) + xlab('Observed values') + ylab('Fitted values') + ggtitle('beatAML, DepInfeR'),
ggplot(data = beatAML_fit) + geom_point(aes(x=Yobs_new, y=GP0_yhat)) +
geom_abline(intercept=0, slope=1) + xlab('Observed values') + ylab('Fitted values') + ggtitle('beatAML, GP'),
ggplot(data = beatAML_fit) + geom_qq(aes(sample=DepInfeR_res)) + geom_qq_line(aes(sample=DepInfeR_res))+
ggtitle('beatAML, DepInfeR') + xlab('Theoretical quantiles') + ylab('Sample quantiles'),
ggplot(data = beatAML_fit) + geom_qq(aes(sample=GP0_res)) + geom_qq_line(aes(sample=GP0_res))+
ggtitle('beatAML, GP') + xlab('Theoretical quantiles') + ylab('Sample quantiles')
)
# EMBL:
load('EMBL_dataset.RData')
load('EMBL_CV.RData')
EMBL_X_new <- tarMat_EMBL
EMBL_X_nan <- EMBL_X_new[1,1]
EMBL_X_new[EMBL_X_new == EMBL_X_nan] <- NaN
EMBL_Y_new <- viabMat_EMBL_log
EMBL_X_original <- tarMat_EMBL
EMBL_Y_original <- viabMat_EMBL
if (!file.exists('EMBL_GP0_inference.RData')){
set.seed(314159)
GP_0_z <- my_gibbs_sampler_gp_0(N=120, burn_in=20, thin=1, X=EMBL_X_new, Y=EMBL_Y_new,
inclusion_prob=0.1,
nu=CV_gp_0_log$c0[[which.min(CV_gp_0_log$mse)]],
update_Z=TRUE)
GP_0_para <- GP_para_0(burn_in=20, X=EMBL_X_new, Y=EMBL_Y_new, Z=GP_0_z$Z,
sigma2=GP_0_z$sigma2, gamma2=GP_0_z$gamma2,
nu=CV_gp_0_log$c0[[which.min(CV_gp_0_log$mse)]])
save(GP_0_z, GP_0_para, file = 'EMBL_GP0_inference.RData')
} else{
load('EMBL_GP0_inference.RData')
}
EMBL_ACAD11 <- mean_fps('ACAD11', EMBL_X_new, GP_0_para)
EMBL_STK26 <- mean_fps('STK26', EMBL_X_new, GP_0_para)
EMBL_CCNT1 <- mean_fps('CCNT1', EMBL_X_new, GP_0_para)
set.seed(314159)
if (!file.exists('DepInfeR_EMBL_res.RData')){
EMBL_DepInfeR_yhat <- MultiLasso(TargetMatrix=EMBL_X_original, ResponseMatrix=EMBL_Y_original, repeats = 100, BPPARAM = bpparam())
} else {load('DepInfeR_EMBL_res.RData')}
EMBL_DepInfeR_Y_hat <- t(t(EMBL_X_original %*% EMBL_DepInfeR_yhat$coefMat) + EMBL_DepInfeR_yhat$intercept)
EMBL_DepInfeR_res <- EMBL_Y_original - t(t(EMBL_X_original %*% EMBL_DepInfeR_yhat$coefMat) + EMBL_DepInfeR_yhat$intercept)
EMBL_GP0_res <- as.vector(EMBL_Y_new - apply(GP_0_para$y_hat, c(1,2), mean))
EMBL_GP0_Y_hat <- as.vector(apply(GP_0_para$y_hat, c(1,2), mean))
EMBL_obs_new <- as.vector(EMBL_Y_new)
EMBL_obs_old <- as.vector(EMBL_Y_original)
EMBL_fit <- data.frame(cbind(as.vector(EMBL_DepInfeR_Y_hat),
as.vector(EMBL_DepInfeR_res),
EMBL_GP0_res, EMBL_GP0_Y_hat, EMBL_obs_new, EMBL_obs_old))
colnames(EMBL_fit) <- c('DepInfeR_yhat', 'DepInfeR_res', 'GP0_res', 'GP0_yhat', 'Yobs_new', 'Yobs_old')
EMBL_fit_fig <- list(ggplot(data = EMBL_fit) + geom_point(aes(x=Yobs_old, y=DepInfeR_yhat)) +
geom_abline(intercept=0, slope=1) + xlab('Observed values') + ylab('Fitted values') + ggtitle('EMBL, DepInfeR'),
ggplot(data = EMBL_fit) + geom_point(aes(x=Yobs_new, y=GP0_yhat)) +
geom_abline(intercept=0, slope=1) + xlab('Observed values') + ylab('Fitted values') + ggtitle('EMBL, GP'),
ggplot(data = EMBL_fit) + geom_qq(aes(sample=DepInfeR_res)) + geom_qq_line(aes(sample=DepInfeR_res))+
ggtitle('EMBL, DepInfeR') + xlab('Theoretical quantiles') + ylab('Sample quantiles'),
ggplot(data = EMBL_fit) + geom_qq(aes(sample=GP0_res)) + geom_qq_line(aes(sample=GP0_res))+
ggtitle('EMBL, GP') + xlab('Theoretical quantiles') + ylab('Sample quantiles')
)
# GDSC1:
load('GDSC1_GP0_inference.RData')
load('GDSC1_dataset.RData')
GDSC1_X_new <- tarMat_GDSC
GDSC1_X_nan <- GDSC1_X_new[1,1]
GDSC1_X_new[GDSC1_X_new == GDSC1_X_nan] <- NaN
GDSC1_Y_new <- viabMat_GDSC_logit
GDSC1_X_original <- tarMat_GDSC
GDSC1_Y_original <- viabMat_GDSC
if (!file.exists('GDSC1_GP0_inference.RData')){
GP_0_z <- my_gibbs_sampler_gp_0(N=120, burn_in=20, thin=1, X=GDSC1_X_new, Y=GDSC1_Y_new,
inclusion_prob=0.1,
nu=CV_gp_0_logit$c0[[which.min(CV_gp_0_logit$mse)]],
update_Z=TRUE)
GP_0_para <- GP_para_0(burn_in=20, X=GDSC1_X_new, Y=GDSC1_Y_new, Z=GP_0_z$Z,
sigma2=GP_0_z$sigma2, gamma2=GP_0_z$gamma2,
nu=CV_gp_0_logit$c0[[which.min(CV_gp_0_logit$mse)]])
save(GP_0_z, GP_0_para, file = 'GDSC1_GP0_inference.RData')
} else{
load('GDSC1_GP0_inference.RData')
}
GDSC1_CCNK <- mean_fps('CCNK', GDSC1_X_new, GP_0_para)
GDSC1_DYRK1A <- mean_fps('DYRK1A', GDSC1_X_new, GP_0_para)
GDSC1_AFF4 <- mean_fps('AFF4', GDSC1_X_new, GP_0_para)
GDSC1_MAP2K2 <- mean_fps('MAP2K2', GDSC1_X_new, GP_0_para)
plot_data <- list(GDSC1_CCNK, GDSC1_DYRK1A, GDSC1_AFF4, GDSC1_MAP2K2,
beatAML_CDK6, beatAML_PDXK, beatAML_AKT1, beatAML_CDK17,
EMBL_ACAD11, EMBL_STK26, EMBL_CCNT1)
kinases <- c("CCNK", "DYRK1A", "AFF4", "MAP2K2",
"CDK6", "PDXK", "AKT1", "CDK17",
"ACAD11", "STK26", "CCNT1")
dataset_name <- c(rep("GDSC1", 4), rep("beatAML", 4), rep('EMBL', 3))
color_vec <- rep(c('black', 'black', 'navyblue', 'navyblue'), 4)
p <- lapply(1:11, function(i){ggplot(data = plot_data[[i]]) +
geom_line(aes(x=Affinity, y=Contribution, group=Sample), alpha=0.4, color = color_vec[i])+
ggtitle(paste0(dataset_name[i], ', kinase=', kinases[i]))})
grid.arrange(grobs=p, ncol=4)
set.seed(314159)
if (!file.exists('DepInfeR_GDSC1_res.RData')){
GDSC1_DepInfeR_yhat <- MultiLasso(TargetMatrix=GDSC1_X_original, ResponseMatrix=GDSC1_Y_original, repeats = 100, BPPARAM = bpparam())
} else {load('DepInfeR_GDSC1_res.RData')}
GDSC1_DepInfeR_Y_hat <- t(t(GDSC1_X_original %*% GDSC1_DepInfeR_yhat$coefMat) + GDSC1_DepInfeR_yhat$intercept)
GDSC1_DepInfeR_res <- GDSC1_Y_original - t(t(GDSC1_X_original %*% GDSC1_DepInfeR_yhat$coefMat) + GDSC1_DepInfeR_yhat$intercept)
GDSC1_GP0_res <- as.vector(GDSC1_Y_new - apply(GP_0_para$y_hat, c(1,2), mean))
GDSC1_GP0_Y_hat <- as.vector(apply(GP_0_para$y_hat, c(1,2), mean))
GDSC1_obs_new <- as.vector(GDSC1_Y_new)
GDSC1_obs_old <- as.vector(GDSC1_Y_original)
GDSC1_fit <- data.frame(cbind(as.vector(GDSC1_DepInfeR_Y_hat),
as.vector(GDSC1_DepInfeR_res),
GDSC1_GP0_res, GDSC1_GP0_Y_hat, GDSC1_obs_new, GDSC1_obs_old))
colnames(GDSC1_fit) <- c('DepInfeR_yhat', 'DepInfeR_res', 'GP0_res', 'GP0_yhat', 'Yobs_new', 'Yobs_old')
GDSC1_fit_fig <- list(ggplot(data = GDSC1_fit) + geom_point(aes(x=Yobs_old, y=DepInfeR_yhat)) +
geom_abline(intercept=0, slope=1) + xlab('Observed values') + ylab('Fitted values') + ggtitle('GDSC1, DepInfeR'),
ggplot(data = subset(GDSC1_fit, !is.na(Yobs_new))) + geom_point(aes(x=Yobs_new, y=GP0_yhat)) +
geom_abline(intercept=0, slope=1) + xlab('Observed values') + ylab('Fitted values') + ggtitle('GDSC1, GP'),
ggplot(data = GDSC1_fit) + geom_qq(aes(sample=DepInfeR_res)) + geom_qq_line(aes(sample=DepInfeR_res))+
ggtitle('GDSC1, DepInfeR') + xlab('Theoretical quantiles') + ylab('Sample quantiles'),
ggplot(data = subset(GDSC1_fit, !is.na(GP0_res))) + geom_qq(aes(sample=GP0_res)) + geom_qq_line(aes(sample=GP0_res))+
ggtitle('GDSC1, GP') + xlab('Theoretical quantiles') + ylab('Sample quantiles')
)
grid.arrange(grobs=c(GDSC1_fit_fig, beatAML_fit_fig, EMBL_fit_fig), ncol=4)