-
Notifications
You must be signed in to change notification settings - Fork 11
/
Copy pathsachs_analysis.R
242 lines (199 loc) · 6.55 KB
/
sachs_analysis.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
# Data files are stored as Excel Spreadsheets with columns as indicated in the headers. The stimulations used are as indicated in the Materials and Methods.
#
#
# For the main result, the following conditions were used:
#
# Stimulation: File .
# 1. CD3, CD28 cd3cd28.xls
# 2. CD3, CD28, ICAM-2 cd3cd28icam2.xls
# 3. CD3, CD28, akt-inhibitor cd3cd28+aktinhib.xls
# 4. CD3, CD28, G0076 cd3cd28+g0076.xls
# 5. CD3, CD28, Psitectorigenin cd3cd28+psitect.xls
# 6. CD3, CD28, U0126 cd3cd28+u0126.xls
# 7. CD3, CD28, LY294002 cd3cd28+ly.xls
# 8. PMA pma.xls
# 9. beta2camp b2camp.xls
#
#
#
# For the simulated western blots, the following were ALSO used, in addition to the 9 above:
#
# Stimulation: File .
# 10. CD3, CD28, ICAM-2, akt-inhib cd3cd28icam2+aktinhib.xls
# 11. CD3, CD28, ICAM-2, G0076 cd3cd28icam2+g0076.xls
# 12. CD3, CD28, ICAM-2, Psitectorigenin cd3cd28icam2+psit.xls
# 13. CD3, CD28, ICAM-2, U0126 cd3cd28icam2+u0126.xls
# 14. CD3, CD28, ICAM-2, LY294002 cd3cd28icam2+ly.xls
####
## Load Sachs data
####
source("source.R")
source("analysis_source.R")
library(readxl)
sachs0 = read.table("bnlearn_files/sachs.data.txt", header = TRUE)
sachs_ints <- list()
sachs_ints[[1]] <- readxl::read_excel("sachs_data/1. cd3cd28.xls")
sachs_ints[[2]] <- readxl::read_excel("sachs_data/2. cd3cd28icam2.xls")
sachs_ints[[3]] <- readxl::read_excel("sachs_data/3. cd3cd28+aktinhib.xls")
sachs_ints[[4]] <- readxl::read_excel("sachs_data/4. cd3cd28+g0076.xls")
sachs_ints[[5]] <- readxl::read_excel("sachs_data/5. cd3cd28+psitect.xls")
sachs_ints[[6]] <- readxl::read_excel("sachs_data/6. cd3cd28+u0126.xls")
sachs_ints[[7]] <- readxl::read_excel("sachs_data/7. cd3cd28+ly.xls")
sachs_ints[[8]] <- readxl::read_excel("sachs_data/8. pma.xls")
sachs_ints[[9]] <- readxl::read_excel("sachs_data/9. b2camp.xls")
head(sachs0)
head(sachs_ints[[1]])
nms <- c("Raf", "Mek", "PLCg", "PIP2", "PIP3", "Erk", "Akt", "PKA", "PKC", "p38", "Jnk")
colnames(sachs0) <- nms
for (i in 1:9) colnames(sachs_ints[[i]]) <- nms
###
# Marginal histograms
###
sachs <- c(list(sachs0), sachs_ints)
layout(matrix(1:12, 4, 3))
par("mar") # 5.1 4.1 4.1 2.1
par(mar = c(2, 2, 2, 2))
for (i in 1:10) hist(log(sachs[[i]]$Raf), main = paste(i))
layout(matrix(1:12, 4, 3))
for (i in 1:10) hist(log(sachs[[i]]$Mek), main = paste(i))
layout(matrix(1:12, 4, 3))
for (i in 1:10) hist(log(sachs[[i]]$PLCg), main = paste(i))
sachs_all <- do.call(rbind, sachs)
## Pairs plot
png("images/sachs_analysis_pairs2.png", width=1000, height=1000)
pairs(log(sachs_all), pch = ".", col = rgb(0,0,0,0.1))
dev.off()
####
## DIAGRAMS GROUP 1: Effect of interventions
####
###
# Raf vs Erk
# C1 == C3 != C6
# Not a good example since Raf also changes
###
layout(1)
yl <- c(0, 6); xl <- c(0, 9)
par(bg = "white")
plot1(log(Erk) ~ log(Raf), sachs_ints[[1]], main = "Control")
plot1(log(Erk) ~ log(Raf), sachs_ints[[3]], main = "Akt-")
plot1(log(Erk) ~ log(Raf), sachs_ints[[6]], main = "Mek-")
triplot(log(Erk) ~ log(Raf), sachs_ints[c(1,3,6)])
###
# PIP3 vs PIP2
# C1 == C3 != C5
# Ok!
###
yl <- c(0, 8); xl <- c(0, 8)
par(bg = "white")
fmla <- log(PIP2) ~ log(PIP3)
plot1(fmla, sachs_ints[[1]], main = "Control")
plot1(fmla, sachs_ints[[3]], main = "Akt-")
plot1(fmla, sachs_ints[[5]], main = "PIP2-")
triplot(fmla, sachs_ints[c(1,3,5)])
###
# PKC vs Akt
# C0 =? C8 =! C9
# OK!
###
yl <- c(0, 9); xl <- c(0, 7)
par(bg = "white")
fmla <- log(Akt) ~ log(PKC)
plot1(fmla, sachs[[1+0]], main = "Control")
plot1(fmla, sachs[[1+8]], main = "PKC+")
plot1(fmla, sachs[[1+9]], main = "PKA+")
triplot(fmla, sachs[1 + c(0,8,9)])
####
## DIAGRAMS: CONDITIONAL INDEPENDENCE PLOTS
####
###
# p38 Jnk | PKA, PKC
###
zattach(sachs_all)
par(bg = "white")
fm <- log(p38) ~ log(Jnk); control <- c("PKA", "PKC"); cf <- log(PKA) ~ log(PKC)
layout(1); par(mar = c(5.1, 4.1, 4.1, 2.1))
layout(1)
marginal_plot()
## subplots
layout(matrix(1:3, 1, 3))
w1 <- getwind()
condition_w(w1)
condition_w(w1, marginal = TRUE)
####
## DIAGRAMS: INVARIANCE
####
###
# Akt | PKA, Erk invariant under perturbation C1 vs C2 vs C3
# Akt | PKA, Erk NOT invariant under perturbation C1 vs C6
###
layout(1)
fmla <- log(Akt) ~ log(PKA) + log(Erk)
dat1 <- subsample(sachs_ints[[1]], 0.5)
dat2 <- subsample(sachs_ints[[3]], 0.5)
yX1 <- model.frame(fmla, data = dat1)
yX2 <- model.frame(fmla, data = dat2)
y1 <- yX1[, 1]
y2 <- yX2[, 1]
X1 <- yX1[, -1]
X2 <- yX2[, -1]
# Xa1 <- data.frame(cbind(y = y1, quadd(X1)))
# Xa2 <- data.frame(cbind(y = y2, quadd(X2)))
Xa1 <- data.frame(cbind(y = y1, X1))
Xa2 <- data.frame(cbind(y = y2, X2))
res1 <- lm(y ~ ., data = Xa1)
res2 <- lm(y ~ ., data = Xa2)
yh11 <- predict(res1, Xa1)
yh21 <- predict(res1, Xa2)
yh12 <- predict(res2, Xa1)
yh22 <- predict(res2, Xa2)
r11 <- y1 - yh11; r12 <- y1 - yh12
r21 <- y2 - yh21; r22 <- y2 - yh22
layout(matrix(1:2, 1, 2))
plot(yh11, yh12); abline(0, 1, col = "red")
plot(yh21, yh22); abline(0, 1, col = "red")
layout(matrix(1:2, 1, 2))
plot(r11, r12); abline(0, 1, col = "red")
plot(r21, r22); abline(0, 1, col = "red")
###
# Mek | Raf, Erk, PKA NOT invariant under perturbation C2, C4
###
layout(1)
fmla <- log(Raf) ~ log(Mek) + log(Erk) + log(PKA)
summary(lm(fmla, data = sachs_all))
summary(lm(fmla, data = sachs_ints[[1]]))
summary(lm(fmla, data = sachs_ints[[2]]))
summary(lm(fmla, data = sachs_ints[[3]]))
summary(lm(fmla, data = sachs_ints[[4]]))
summary(lm(fmla, data = sachs_ints[[5]]))
summary(lm(fmla, data = sachs_ints[[6]]))
summary(lm(fmla, data = sachs_ints[[7]]))
summary(lm(fmla, data = sachs_ints[[8]]))
summary(lm(fmla, data = sachs_ints[[9]]))
# pairs(sachs_ints[[1]], pch = ".")
dat1 <- subsample(sachs_ints[[2]], 0.5)
dat2 <- subsample(sachs_ints[[4]], 0.5)
# dat1 <- sachs_ints[[1]]
yX1 <- model.frame(fmla, data = dat1)
yX2 <- model.frame(fmla, data = dat2)
y1 <- yX1[, 1]
y2 <- yX2[, 1]
X1 <- yX1[, -1]
X2 <- yX2[, -1]
Xa1 <- data.frame(cbind(y = y1, quadd(X1)))
Xa2 <- data.frame(cbind(y = y2, quadd(X2)))
# Xa1 <- data.frame(cbind(y = y1, X1))
# Xa2 <- data.frame(cbind(y = y2, X2))
res1 <- lm(y ~ ., data = Xa1)
res2 <- lm(y ~ ., data = Xa2)
yh11 <- predict(res1, Xa1)
yh21 <- predict(res1, Xa2)
yh12 <- predict(res2, Xa1)
yh22 <- predict(res2, Xa2)
r11 <- y1 - yh11; r12 <- y1 - yh12
r21 <- y2 - yh21; r22 <- y2 - yh22
layout(matrix(1:2, 1, 2))
plot(yh11, yh12); abline(0, 1, col = "red")
plot(yh21, yh22); abline(0, 1, col = "red")
layout(matrix(1:2, 1, 2))
plot(r11, r12); abline(0, 1, col = "red")
plot(r21, r22); abline(0, 1, col = "red")