-
Notifications
You must be signed in to change notification settings - Fork 1
/
Analysis3_DE.Rmd
356 lines (284 loc) · 12.4 KB
/
Analysis3_DE.Rmd
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
---
title: 'Result Part 3: Differential Expression Analysis'
output: html_notebook
---
# Differential Gene Analysis
- datasets from
+ Segerstolpe et al., RPKM data, 4 T2D patients and 3 healthy donors
+ Chu celltype dataset comparing H1 and DE cells
- use DE (t-test) on Bulk RNA-seq as ground truth
- plot AUC of DE (t-test, Wilcox) on scRNA-seq against ground truth
- DE analysis performed using Seurat 3.0
##1. Segersope DE analysis
### Raw data preprocessing
- filter out genes w. zero proportion > 80%
```{r}
load("Data/T2D.rda")
library(Seurat)
summarize <- function(df){
zero_prop = apply(df, 1, function(x)sum(x==0, na.rm = T))/ncol(df)
cat("* summary of zero proportion\n")
print(summary(zero_prop))
lib_size = apply(df,2, function(x)sum(x, na.rm = T))
cat("* summary of library size\n")
print(summary(lib_size))
return(list(zero_prop = zero_prop,lib_size = lib_size))
}
summary.list = summarize(as.matrix(raw.rpkm))
gene.fil <- summary.list$zero_prop<0.80
raw.rpkm.fil <- raw.rpkm[gene.fil,]
save(raw.rpkm.fil, file ='Data/T2D_raw_fil.rda' )
```
### Read in imputed datasets
```{r}
load("Data/T2D.rda")
load('Data/T2D_raw_fil.rda')
raw.rpkm.scimpute <- readRDS('Data/seger_raw_rpkm_scimpute.rds')
raw.rpkm.g2s3 <- readRDS("Data/seger_raw_rpkm_g2s3.rds")
raw.rpkm.dca = readRDS("Data/seger_raw_rpkm_dca.rds")
raw.rpkm.saver = readRDS('Data/T2D_raw_fil_saver.rds')$estimate
raw.rpkm.magic = readRDS('Data/seger_raw_rpkm_magic.rds')
#add new methods
raw.rpkm.sctssr = readRDS("Data/seger_raw_rpkm_sctssr.rds")
raw.rpkm.alra = readRDS("Data/seger_raw_rpkm_alra.rds")
raw.rpkm.knnsmooth = readRDS("Data/seger_raw_rpkm_knnsmooth.rds")
raw.rpkm.saucie = readRDS("Data/seger_raw_rpkm_saucie.rds")
raw.rpkm.enimpute = readRDS("Data/seger_raw_rpkm_enimpute.rds")
dimnames(raw.rpkm.enimpute)=dimnames(raw.rpkm.alra)
#select 7 subjects w. both bulk and scRNA
subjectID <- unlist(lapply(strsplit(colnames(bulk.rpkm)[-1], "_"), "[[", 1))
sc_col.sel <- grepl(paste(subjectID, collapse="|"), colnames(raw.rpkm.scimpute))
sc.list = list(bulk = bulk.rpkm[rownames(raw.rpkm.fil),-1],
raw = raw.rpkm.fil[,sc_col.sel],
g2s3 = raw.rpkm.g2s3[,sc_col.sel],
scimpute = raw.rpkm.scimpute[,sc_col.sel],
saver = raw.rpkm.saver[, sc_col.sel],
dca = raw.rpkm.dca[, sc_col.sel],
magic = raw.rpkm.magic[, sc_col.sel],
sctssr = raw.rpkm.sctssr[, sc_col.sel],
# saucie = raw.rpkm.saucie[, sc_col.sel],
knnsmooth = raw.rpkm.knnsmooth[, sc_col.sel],
alra = raw.rpkm.alra[, sc_col.sel],
enimpute = raw.rpkm.enimpute[, sc_col.sel])
lapply(sc.list, function(x) head(colnames(x)))
lapply(sc.list, function(x) head(rownames(x)))
save(sc.list, file = "Data/DE_rpkm_list_final.rda")
```
### DE analysis
```{r}
load("Data/DE_rpkm_list_final.rda")
library(Seurat)
# DE test with t-test on bulk
DE_test <- function(x, de_method){
df <- CreateSeuratObject(counts = x)
stim <- rep("ctrl", ncol(df)); stim[grep("T2D", colnames(df))] <- "stim"
df$stim <- stim
df <- subset(df, subset=nFeature_RNA>500)
df <- FindVariableFeatures(df, selection.method = "vst", nfeatures = 5000)
df <- NormalizeData(df, normalization.method = "LogNormalize", scale.factor = 10000)
Idents(df) <- "stim"
if (ncol(df)<10) de_method = 't'
de.markers <- FindMarkers(df,ident.1 = "stim", ident.2 = "ctrl", test.use = de_method)
# FindMarkers(df,ident.1 = "stim", ident.2 = "ctrl", test.use = de_method)
return(de.markers)
}
library(DESeq2)
# de test with DESeq2 on bulk
DE_test2 <- function(x, de_method){
# use DESeq2 for bulk
x = round(x)
df <- CreateSeuratObject(counts = x)
stim <- rep("ctrl", ncol(df)); stim[grep("T2D", colnames(df))] <- "stim"
df$stim <- stim
df <- subset(df, subset=nFeature_RNA>500)
df <- FindVariableFeatures(df, selection.method = "vst", nfeatures = 5000)
df <- NormalizeData(df, normalization.method = "LogNormalize", scale.factor = 10000)
Idents(df) <- "stim"
if (ncol(df)<10) de_method = 'DESeq2'
de.markers <- FindMarkers(df,ident.1 = "stim", ident.2 = "ctrl", test.use = de_method)
# FindMarkers(df,ident.1 = "stim", ident.2 = "ctrl", test.use = de_method)
return(de.markers)
}
ex_level = apply(sc.list[[2]], 1, function(x) sum(x!=0, na.rm = T))/ncol(sc.list[[2]])
high = (ex_level>0.3); low = (ex_level <=0.3);
sum(high); sum(low)
# Formalize the name
names(sc.list)
# reorder the methods
# c("Raw", "G2S3", "SAVER", "kNN-smoothing", "MAGIC", "scImpute", "VIPER", "ALRA", "scTSSR", "DCA", "SAUCIE", "EnImpute")
sc.list <- sc.list[c(1, 2, 3, 5, 9, 7, 4, 10, 8, 6, 11)]
names(sc.list)
names(sc.list) = c("Bulk", "Raw", "G2S3", "SAVER", "kNN-smoothing", "MAGIC", "scImpute", "ALRA", "scTSSR", "DCA", "EnImpute")
#high
sc.list.high = lapply(sc.list, function(x)x[which(high),])
de.ttest.high <- lapply(sc.list.high, function(x) DE_test(x, 't'))
de.wilcox.high <- lapply(sc.list.high, function(x) DE_test(x, 'wilcox'))
#low (magic removed (No features pass logfc.threshold threshold))
sc.list.low = lapply(sc.list, function(x)x[which(low),])
sc.list.low[[6]] = NULL
names(sc.list.low)
de.ttest.low <- lapply(sc.list.low, function(x) DE_test(x, 't'))
de.wilcox.low <- lapply(sc.list.low, function(x) DE_test(x, 'wilcox'))
de.ttest.all <- lapply(sc.list, function(x) DE_test(x, 't'))
de.wilcox.all <- lapply(sc.list, function(x) DE_test(x, 'wilcox'))
# rpkm data (need round the number to get integers)
de.ttest.deseq <- lapply(sc.list, function(x) DE_test2(x, 't'))
de.wilcox.deseq <- lapply(sc.list, function(x) DE_test2(x,'wilcox'))
save(de.ttest.high, de.ttest.low, de.wilcox.high, de.wilcox.low, de.ttest.all, de.wilcox.all,de.ttest.deseq,de.wilcox.deseq ,file = "Data/de_results_final.rda")
```
### Generate Overall DE plot in Segerstolpe dataset
```{r}
calc_auc = function(data, label){
library(ROCR)
library(tidyverse)
colors = c("black", "#cb181d", "#6baed6", '#8491B4B2', "#bae4b3", 'orange', "#FFD700", '#3399FF', "#66CC99", '#D55E00', '#7570b3')
if (length(data)==10) colors = colors[-7]
methods = names(data)
pred_all <- lapply(data[2:length(methods)], function(x){
common.genes <- intersect(rownames(x), rownames(data[[1]]))
df <- data.frame(cbind(data[[1]][common.genes, "p_val"], x[common.genes, 'p_val']))
colnames(df) = c("bulk", "imp")
df$true_label = 0;df$true_label[df$bulk<0.05] = 1
pred <- prediction(-log(df$imp), df$true_label)
return(pred)
})
auc_all <- lapply(pred_all, function(x) performance(x, measure = "auc"))
perf_all <- lapply(pred_all, function(x) performance(x, "tpr","fpr"))
AUCs = unlist(unlist(auc_all)%>%lapply("slot", 'y.values'))
for(i in 1:length(perf_all)){
par.add = (i!=1)
par(cex.axis=1.4, cex = 1.4, cex.lab = 1.4, lwd = 3, cex.main = 2)
plot(perf_all[[i]], col=colors[i] , add = par.add, main = "",lwd = 4, xlab = "False Positive Rate", ylab = "True Positive Rate")
print(paste(names(perf_all)[[i]], round(AUCs[i],2)))
}
legend(0.57, 0.45, legend=paste(methods[-1],round(AUCs,2), sep = ": "),
col=colors, lty = 1, cex=1)
abline(a = 0, b = 1)
}
pdf('Plots/Figure_response_overall.pdf',width=22, height=11)
par(mfrow = c(1,2), oma = c(2, 2, 2, 2))
calc_auc(de.ttest.all, '')
text(0.5,
line2user(line=3, side=3), 'T-test', xpd=NA, cex=2, font=2)
calc_auc(de.wilcox.all, '')
text(0.5,
line2user(line=3, side=3), 'Wilcoxon Test', xpd=NA, cex=2, font=2)
dev.off()
pdf('Plots/Figure_response_deseq.pdf',width=22, height=11)
par(mfrow = c(1,2), oma = c(2, 2, 2, 2))
calc_auc(de.ttest.deseq, '')
text(0.5,
line2user(line=3, side=3), 'T-test', xpd=NA, cex=2, font=2)
calc_auc(de.wilcox.deseq, '')
text(0.5,
line2user(line=3, side=3), 'Wilcoxon Test', xpd=NA, cex=2, font=2)
dev.off()
```
##2. Chu's datasets
### DE analysis
```{r}
load("../Clustering/Data/Chu_final.rda")
data_all = data[c(1, c(1, 2, 3, 10, 4, 5, 6, 11, 8, 7, 9, 12)+1)]
names(data)
names(data) <- c("Bulk", "Raw", "G2S3", "SAVER", "kNN-smoothing", "MAGIC", "scImpute", "VIPER", "ALRA", "scTSSR", "DCA", "SAUCIE", "EnImpute")
library(Seurat)
# H1 vs DEC
data = lapply(data_all, function(x) {x = x[,grep("H1_|DEC_", colnames(x))]; return(x)})
DE_test.chu <- function(x, de_method){
df <- CreateSeuratObject(counts = x)
stim <- rep("ctrl", ncol(df)); stim[grep("DEC_", colnames(df))] <- "stim"
df$stim <- stim
df <- subset(df, subset=nFeature_RNA>500)
df <- FindVariableFeatures(df, selection.method = "vst", nfeatures = 5000)
df <- NormalizeData(df, normalization.method = "LogNormalize", scale.factor = 10000)
Idents(df) <- "stim"
if (ncol(df)<100) de_method = 't' # always use t-test for bulk
de.markers <- FindMarkers(df,ident.1 = "stim", ident.2 = "ctrl", test.use = de_method,min.cells.group = 1)
return(de.markers)
}
DE_test.all <- function(x, de_method){
df <- CreateSeuratObject(counts = x)
stim <- sapply(strsplit(colnames(x), "_"), "[", 1)
df$stim <- stim
df <- subset(df, subset=nFeature_RNA>500)
df <- FindVariableFeatures(df, selection.method = "vst", nfeatures = 5000)
df <- NormalizeData(df, normalization.method = "LogNormalize", scale.factor = 10000)
Idents(df) <- "stim"
if (ncol(df)<100) de_method = 't' # always use t-test for bulk
de.markers <- FindAllMarkers(df,test.use = de_method,min.cells.group = 1)
return(de.markers)
}
bulk_de = DE_test.all(data_all[[1]], "t")
data[['SAUCIE']] <- NULL # saucie can't find markers
calc_auc2 = function(res, label){
library(ROCR)
library(tidyverse)
colors = c("black", "#cb181d", "#6baed6", '#8491B4B2', "#bae4b3", 'orange', "#FFD700", '#3399FF', "#66CC99", '#D55E00', '#7570b3')
if (length(res)==10) colors = colors[-7]
methods = names(res)
pred_all <- lapply(res[2:length(methods)], function(x){
common.genes <- intersect(rownames(res[[2]]), rownames(res[[1]]))
df <- data.frame(cbind(res[[1]][common.genes, "p_val_adj"], x[common.genes, 'p_val_adj']))
colnames(df) = c("bulk", "imp")
df$true_label = 0;df$true_label[df$bulk<0.05] = 1
pred <- prediction(-log(df$imp), df$true_label)
return(pred)
})
auc_all <- lapply(pred_all, function(x) performance(x, measure = "auc"))
perf_all <- lapply(pred_all, function(x) performance(x, "tpr","fpr"))
AUCs = unlist(unlist(auc_all)%>%lapply("slot", 'y.values'))
for(i in 1:length(perf_all)){
par.add = (i!=1)
par(cex.axis=1.4, cex = 1.4, cex.lab = 1.4, lwd = 3, cex.main = 2)
plot(perf_all[[i]], col=colors[i] , add = par.add, main = "",lwd = 4, xlab = "False Positive Rate", ylab = "True Positive Rate")
print(paste(names(perf_all)[[i]], round(AUCs[i],2)))
}
legend(0.57, 0.50, legend=paste(methods[-1],round(AUCs,2), sep = ": "),
col=colors, lty = 1, cex=1)
abline(a = 0, b = 1)
}
de.ttest<- lapply(data, function(x) DE_test.chu(x, 't'))
de.wilcox<- lapply(data, function(x) DE_test.chu(x,'wilcox'))
ex_level = apply(data[[2]], 1, function(x) sum(x!=0, na.rm = T))/ncol(data[[2]])
high = (ex_level>0.7); low = (ex_level<=0.7) ;
sum(high); sum(low)
de.ttest.high <- lapply(de.ttest, function(x)x[intersect(names(which(high)),rownames(x)), ] )
de.ttest.low <- lapply(de.ttest, function(x)x[intersect(names(which(low)),rownames(x)), ])
de.wilcox.high <- lapply(de.wilcox, function(x) x[intersect(names(which(high)),rownames(x)),, ])
de.wilcox.low <- lapply(de.wilcox, function(x)x[intersect(names(which(low)),rownames(x)),,])
save(de.ttest.high, de.ttest.low, de.wilcox.high, de.wilcox.low, de.ttest, de.wilcox,file = "Data/de_results_CHU_0.7.rda")
```
##3. Generate Figure 5
```{r}
addfiglab <- function(lab, xl = par()$mar[2], yl = par()$mar[3]) {
text(x = line2user(xl, 2), y = line2user(yl, 3),
lab, xpd = NA, font = 2, cex = 3, adj = c(0, 1))
}
pdf('Plots/Figure5_rev2.pdf',width=40, height=20)
load("Data/de_results_final.rda")
par(mfrow = c(2,4), oma = c(2, 2, 2, 2))
calc_auc(de.ttest.high, '')
addfiglab("A")
calc_auc(de.ttest.low, '')
text(line2user(line=mean(par('mar')[c(2, 4)]), side=2),
line2user(line = 3.5, side=3), 'T-test', xpd=NA, cex=2, font=2)
calc_auc(de.wilcox.high, '')
calc_auc(de.wilcox.low, '')
text(line2user(line=mean(par('mar')[c(2, 4)]), side=2),
line2user(line=3.5, side=3), 'Wilcoxon Test', xpd=NA, cex=2, font=2)
load('Data/de_results_CHU_0.7.rda')
calc_auc2(de.ttest.high, '')
addfiglab("B")
text(0.5,
line2user(line=6, side=1), 'Highly Expressed', xpd=NA, cex=2, font=2)
calc_auc2(de.ttest.low, '')
text(0.5,
line2user(line=6, side=1), 'Lowly Expressed', xpd=NA, cex=2, font=2)
calc_auc2(de.wilcox.high, '')
text(0.5,
line2user(line=6, side=1), 'Highly Expressed', xpd=NA, cex=2, font=2)
calc_auc2(de.wilcox.low, '')
text(0.5,
line2user(line=6, side=1), 'Lowly Expressed', xpd=NA, cex=2, font=2)
dev.off()
```