-
Notifications
You must be signed in to change notification settings - Fork 1
/
cd8_analysis.R
217 lines (175 loc) · 12.8 KB
/
cd8_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
library(Seurat)
library(ggplot2)
library(xlsx)
library(fgsea)
# Enrichment reference set
go.set = gmtPathways('c5.all.v6.1.symbols.gmt')
## READ IN RAW DATA AND META DATA
raw.data = read.table("CD8_rsem.txt",sep = "\t", header = TRUE, row.names=1)
#raw.data=log(raw.data+1)
meta = read.table("cd8_qc2.txt",sep = "\t", header = TRUE, row.names = 1)
## SET UP SEURAT OBJECT
cd8 = CreateSeuratObject(raw.data=raw.data,project = "cd8",min.cells = 19,names.field=1,min.genes=1000,is.expr=0,meta.data=meta, do.scale=FALSE)
cd8 = NormalizeData(object= cd8, normalization.method="LogNormalize")
cd8 = FindVariableGenes(object = cd8,mean.function = ExpMean,dispersion.function=LogVMR, x.low.cutoff=0.5,y.cutoff=0,do.plot=TRUE)
cd8 = ScaleData(object=cd8)
cd8 = RunPCA(object=cd8,pcs.compute=15,do.print=TRUE,pcs.print=1:5,genes.print=1:5)
cd8.24 = which(cd8@meta.data$time==24 & cd8@meta.data$mar>-100)
cd8.48 = which(cd8@meta.data$time==48 & cd8@meta.data$mar>-100)
cd8.tot = c(cd8.24,cd8.48)
# Determine differentially expressed genes -- Supp. Tables 5/6
markers = FindMarkers(cd8,ident.1 = 'T24hr', ident.2 = 'T48hr')
# Sort based on average diff. exp. fold change
markers.idx = sort(markers[,2], decreasing=FALSE, index.return=TRUE)$ix
markers.sort = as.numeric(markers[markers.idx,2])
names(markers.sort) = toupper(rownames(markers)[markers.idx])
marker.ranked.enrichment = fgsea(go.set,markers.sort,nperm=15000,maxSize = 500)
marker.ranked.enrichment = marker.ranked.enrichment[sort(marker.ranked.enrichment$NES,index.return=TRUE, decreasing = FALSE)$ix,1:6]
# Figure 4a --> Mass and Mar for both time points
par(pty = "s")
plot(cd8@meta.data$mass[cd8.24],cd8@meta.data$mar[cd8.24], xlim = c(20,80), ylim = c(0,8.5), pch=21, cex=1.75, bg = rgb(0,0,1,1),xlab = "Mass (pg)",ylab = "MAR (pg/h)", cex.axis = 1.5, cex.lab=1.5)
points(cd8@meta.data$mass[cd8.48],cd8@meta.data$mar[cd8.48], xlim = c(20,80), ylim = c(0,8.5), pch=24, cex=1.75, bg = rgb(1,0,0,1))
# Figure 4b --> Plot normalized growth rates
norm.24 = cd8@meta.data$mar[cd8.24]/cd8@meta.data$mass[cd8.24]
norm.48 = cd8@meta.data$mar[cd8.48]/cd8@meta.data$mass[cd8.48]
x1 = rep(1,length(norm.24))
x2 = rep(2,length(norm.48))
par(pty = "s")
plot(jitter(x1,10),norm.24,xlim= c(.5,2.5), ylim = c(0,0.15),bg=rgb(0,0,1,1),pch=21,cex = 1.75, ylab = "Normalized MAR (1/h)",cex.lab = 1.5, cex.axis=1.5)
# Generate supplementary gene lists and enrichment analyses
# Mass -- Pearson
mass.corr.pears = apply(t(as.matrix(cd8@data[,cd8.24])),2,function(x) cor(x,cd8@meta.data$mass[cd8.24]))
mass.corr.pears[is.na(mass.corr.pears)]=0
# Mass -- Spearman
mass.corr.spear.24 = apply(t(as.matrix(cd8@data[,cd8.24])),2,function(x) cor(x,cd8@meta.data$mass[cd8.24],method='spearman'))
mass.corr.spear.24[is.na(mass.corr.spear.24)]=0
mass.corr.spear.24.rank = sort(mass.corr.spear.24, decreasing=TRUE,index.return=TRUE)$ix
mass.ranked.list = mass.corr.spear.24[mass.corr.spear.24.rank]
names(mass.ranked.list) = toupper(names(mass.corr.spear.24)[mass.corr.spear.24.rank])
mass.ranked.enrichment = fgsea(go.set,mass.ranked.list,nperm=10000,maxSize = 500)
mass.ranked.enrichment <- mass.ranked.enrichment[mass.ranked.enrichment$padj< 0.1]
mass.ranked.enrichment$NES[is.na(mass.ranked.enrichment$NES)]=0
# Null Distribution for mass
cd8.mass.null = sapply(1:10, function(x) sample(cd8.24,length(cd8.24),replace=FALSE))
cd8.mass.corr.null = sapply(1:10,function(x) (apply(cd8@data[,cd8.24],1,function(y) cor(y,cd8@meta.data$mass[cd8.mass.null[,x]],method='spearman'))))
cd8.mass.corr.null[is.na(cd8.mass.corr.null)]=0
cd8.mass.null.mean = mean(colMeans(cd8.mass.corr.null))
cd8.mass.null.sd = mean(apply(cd8.mass.corr.null,2,function(x) sd(x)))
cols<- c("blue","red")[(abs(mass.corr.spear.24[mass.corr.spear.24.rank])>2*cd8.mass.null.sd) + 1]
barplot(mass.corr.spear.24[mass.corr.spear.24.rank], col=cols,border=NA,ylim = c(-0.75,.75),xaxt='n', main = c('cd8 24 mass',as.character(length(which(mass.corr.spear.24>2*cd8.mass.null.sd))),as.character(length(which(mass.corr.spear.24< -2*cd8.mass.null.sd)))))
box()
abline(h = 2*cd8.mass.null.sd,lwd=2,col=1,lty=2)
abline(h = -2*cd8.mass.null.sd,lwd=2,col=1,lty=2)
abline(h=0,lwd=1)
# Plot Spearman v. Pearson for cd8 mass correlations
plot(mass.corr.spear.24, mass.corr.pears, pch = 21, bg = rgb(0,0,1,0.3), xlab = 'Spearman Coefficient', ylab = 'Pearson Coefficeint', xlim = c(-0.8,0.8), ylim = c(-0.8,.8), cex.lab =1.2, cex.axis = 1.2,cex=0.75,col=rgb(0,0,0,0))
title(c('cd8 24 Mass', as.character(cor(mass.corr.spear.24,mass.corr.pears))))
# Normalized growth rate 24h
cd8@meta.data$norm = cd8@meta.data$mar/cd8@meta.data$mass
# norm -- Pearson
norm.corr.pears = apply(t(as.matrix(cd8@data[,cd8.24])),2,function(x) cor(x,cd8@meta.data$norm[cd8.24]))
norm.corr.pears[is.na(norm.corr.pears)]=0
# norm -- Spearman
norm.corr.spear.24 = apply(t(as.matrix(cd8@data[,cd8.24])),2,function(x) cor(x,cd8@meta.data$norm[cd8.24],method='spearman'))
norm.corr.spear.24[is.na(norm.corr.spear.24)]=0
norm.corr.spear.24.rank = sort(norm.corr.spear.24, decreasing=TRUE,index.return=TRUE)$ix
norm.ranked.list = norm.corr.spear.24[norm.corr.spear.24.rank]
names(norm.ranked.list) = toupper(names(norm.corr.spear.24)[norm.corr.spear.24.rank])
norm.ranked.enrichment = fgsea(go.set,norm.ranked.list,nperm=10000,maxSize = 500)
norm.ranked.enrichment <- norm.ranked.enrichment[norm.ranked.enrichment$padj< 0.1]
norm.ranked.enrichment$NES[is.na(norm.ranked.enrichment$NES)]=0
norm.ranked.enrichment = norm.ranked.enrichment[sort(norm.ranked.enrichment$NES,index.return=TRUE, decreasing = FALSE)$ix,1:6]
# Null Distribution for norm
cd8.norm.null = sapply(1:10, function(x) sample(cd8.24,length(cd8.24),replace=FALSE))
cd8.norm.corr.null = sapply(1:10,function(x) (apply(cd8@data[,cd8.24],1,function(y) cor(y,cd8@meta.data$norm[cd8.norm.null[,x]],method='spearman'))))
cd8.norm.corr.null[is.na(cd8.norm.corr.null)]=0
cd8.norm.null.mean = mean(colMeans(cd8.norm.corr.null))
cd8.norm.null.sd = mean(apply(cd8.norm.corr.null,2,function(x) sd(x)))
cols<- c("blue","red")[(abs(norm.corr.spear.24[norm.corr.spear.24.rank])>2*cd8.norm.null.sd) + 1]
barplot(norm.corr.spear.24[norm.corr.spear.24.rank], col=cols,border=NA,ylim = c(-0.75,.75),xaxt='n', main = c('cd8 24 norm',as.character(length(which(norm.corr.spear.24>2*cd8.norm.null.sd))),as.character(length(which(norm.corr.spear.24< -2*cd8.norm.null.sd)))))
box()
abline(h = 2*cd8.norm.null.sd,lwd=2,col=1,lty=2)
abline(h = -2*cd8.norm.null.sd,lwd=2,col=1,lty=2)
abline(h=0,lwd=1)
# Plot Spearman v. Pearson for cd8 norm correlations
plot(norm.corr.spear.24, norm.corr.pears, pch = 21, bg = rgb(0,0,1,0.3), xlab = 'Spearman Coefficient', ylab = 'Pearson Coefficeint', xlim = c(-0.8,0.8), ylim = c(-0.8,.8), cex.lab =1.2, cex.axis = 1.2,cex=0.75,col=rgb(0,0,0,0))
title(c('cd8 24 norm', as.character(cor(norm.corr.spear.24,norm.corr.pears))))
# 48h analysis
# Mass -- Pearson
mass.corr.pears = apply(t(as.matrix(cd8@data[,cd8.48])),2,function(x) cor(x,cd8@meta.data$mass[cd8.48]))
mass.corr.pears[is.na(mass.corr.pears)]=0
# Mass -- Spearman
mass.corr.spear.48 = apply(t(as.matrix(cd8@data[,cd8.48])),2,function(x) cor(x,cd8@meta.data$mass[cd8.48],method='spearman'))
mass.corr.spear.48[is.na(mass.corr.spear.48)]=0
mass.corr.spear.48.rank = sort(mass.corr.spear.48, decreasing=TRUE,index.return=TRUE)$ix
mass.ranked.list = mass.corr.spear.48[mass.corr.spear.48.rank]
names(mass.ranked.list) = toupper(names(mass.corr.spear.48)[mass.corr.spear.48.rank])
mass.ranked.enrichment = fgsea(go.set,mass.ranked.list,nperm=10000,maxSize = 500)
mass.ranked.enrichment <- mass.ranked.enrichment[mass.ranked.enrichment$padj< 0.1]
mass.ranked.enrichment$NES[is.na(mass.ranked.enrichment$NES)]=0
mass.ranked.enrichment = mass.ranked.enrichment[sort(mass.ranked.enrichment$NES,index.return=TRUE, decreasing = FALSE)$ix,1:6]
# Null Distribution for mass
cd8.mass.null = sapply(1:10, function(x) sample(cd8.48,length(cd8.48),replace=FALSE))
cd8.mass.corr.null = sapply(1:10,function(x) (apply(cd8@data[,cd8.48],1,function(y) cor(y,cd8@meta.data$mass[cd8.mass.null[,x]],method='spearman'))))
cd8.mass.corr.null[is.na(cd8.mass.corr.null)]=0
cd8.mass.null.mean = mean(colMeans(cd8.mass.corr.null))
cd8.mass.null.sd = mean(apply(cd8.mass.corr.null,2,function(x) sd(x)))
cols<- c("blue","red")[(abs(mass.corr.spear.48[mass.corr.spear.48.rank])>2*cd8.mass.null.sd) + 1]
barplot(mass.corr.spear.48[mass.corr.spear.48.rank], col=cols,border=NA,ylim = c(-0.75,.75),xaxt='n', main = c('cd8 48 mass',as.character(length(which(mass.corr.spear.48>2*cd8.mass.null.sd))),as.character(length(which(mass.corr.spear.48< -2*cd8.mass.null.sd)))))
box()
abline(h = 2*cd8.mass.null.sd,lwd=2,col=1,lty=2)
abline(h = -2*cd8.mass.null.sd,lwd=2,col=1,lty=2)
abline(h=0,lwd=1)
# Plot Spearman v. Pearson for cd8 mass correlations
plot(mass.corr.spear.48, mass.corr.pears, pch = 21, bg = rgb(0,0,1,0.3), xlab = 'Spearman Coefficient', ylab = 'Pearson Coefficeint', xlim = c(-0.8,0.8), ylim = c(-0.8,.8), cex.lab =1.2, cex.axis = 1.2,cex=0.75,col=rgb(0,0,0,0))
title(c('cd8 48 Mass', as.character(cor(mass.corr.spear.48,mass.corr.pears))))
# Normalized growth rate 48h
cd8@meta.data$norm = cd8@meta.data$mar/cd8@meta.data$mass
# norm -- Pearson
norm.corr.pears = apply(t(as.matrix(cd8@data[,cd8.48])),2,function(x) cor(x,cd8@meta.data$norm[cd8.48]))
norm.corr.pears[is.na(norm.corr.pears)]=0
# norm -- Spearman
norm.corr.spear.48 = apply(t(as.matrix(cd8@data[,cd8.48])),2,function(x) cor(x,cd8@meta.data$norm[cd8.48],method='spearman'))
norm.corr.spear.48[is.na(norm.corr.spear.48)]=0
norm.corr.spear.48.rank = sort(norm.corr.spear.48, decreasing=TRUE,index.return=TRUE)$ix
norm.ranked.list = norm.corr.spear.48[norm.corr.spear.48.rank]
names(norm.ranked.list) = toupper(names(norm.corr.spear.48)[norm.corr.spear.48.rank])
norm.ranked.enrichment = fgsea(go.set,norm.ranked.list,nperm=10000,maxSize = 500)
norm.ranked.enrichment <- norm.ranked.enrichment[norm.ranked.enrichment$padj< 0.1]
norm.ranked.enrichment$NES[is.na(norm.ranked.enrichment$NES)]=0
norm.ranked.enrichment = norm.ranked.enrichment[sort(norm.ranked.enrichment$NES,index.return=TRUE, decreasing = FALSE)$ix,1:6]
# Null Distribution for norm
cd8.norm.null = sapply(1:10, function(x) sample(cd8.48,length(cd8.48),replace=FALSE))
cd8.norm.corr.null = sapply(1:10,function(x) (apply(cd8@data[,cd8.48],1,function(y) cor(y,cd8@meta.data$norm[cd8.norm.null[,x]],method='spearman'))))
cd8.norm.corr.null[is.na(cd8.norm.corr.null)]=0
cd8.norm.null.mean = mean(colMeans(cd8.norm.corr.null))
cd8.norm.null.sd = mean(apply(cd8.norm.corr.null,2,function(x) sd(x)))
cols<- c("blue","red")[(abs(norm.corr.spear.48[norm.corr.spear.48.rank])>2*cd8.norm.null.sd) + 1]
barplot(norm.corr.spear.48[norm.corr.spear.48.rank], col=cols,border=NA,ylim = c(-0.75,.75),xaxt='n', main = c('cd8 48 norm',as.character(length(which(norm.corr.spear.48>2*cd8.norm.null.sd))),as.character(length(which(norm.corr.spear.48< -2*cd8.norm.null.sd)))))
box()
abline(h = 2*cd8.norm.null.sd,lwd=2,col=1,lty=2)
abline(h = -2*cd8.norm.null.sd,lwd=2,col=1,lty=2)
abline(h=0,lwd=1)
# Plot Spearman v. Pearson for cd8 norm correlations
plot(norm.corr.spear.48, norm.corr.pears, pch = 21, bg = rgb(0,0,1,0.3), xlab = 'Spearman Coefficient', ylab = 'Pearson Coefficeint', xlim = c(-0.8,0.8), ylim = c(-0.8,.8), cex.lab =1.2, cex.axis = 1.2,cex=0.75,col=rgb(0,0,0,0))
title(c('cd8 48 norm', as.character(cor(norm.corr.spear.48,norm.corr.pears))))
# Figure 3c -- Correlation with genes from Kimmerling et al., Nat. Comm. 2016
cc.genes = as.matrix(read.table("Cell_cycle_CD8.txt",sep="\t",header = TRUE))
cc.idx = match(cc.genes,rownames(cd8@data))
cc.idx = cc.idx[!is.na(cc.idx)]
# Create null distributions for correlation coefficients (limited to 10 iterations to reduce run time)
cd8.24.null.idx = sapply(1:10,function(x) sample(cd8.24,length(cd8.24),replace=FALSE))
null.mat.24 = sapply(1:10,function(x) (apply(cd8@data[cc.idx,cd8.24],1,function(y) cor(y,cd8@meta.data$mass[cd8.24.null.idx[,x]]))))
null.mean.24 = mean(colMeans(null.mat.24))
null.sd.24 = mean(apply(null.mat.24,2,function(x) sd(x)))
null.vec.24 = rnorm(10,mean=null.mean.24,sd=null.sd.24)
# 48h
cd8.48.null.idx = sapply(1:10,function(x) sample(cd8.48,length(cd8.48),replace=FALSE))
null.mat.48 = sapply(1:10,function(x) (apply(cd8@data[cc.idx,cd8.48],1,function(y) cor(y,cd8@meta.data$mass[cd8.48.null.idx[,x]]))))
null.mean.48 = mean(colMeans(null.mat.48))
null.sd.48 = mean(apply(null.mat.48,2,function(x) sd(x)))
null.vec.48 = rnorm(10,mean=null.mean.48,sd=null.sd.48)
# Find correlations between mass and cluster genes for both time points
cl3.corr.24 = apply(cd8@data[cc.idx,cd8.24],1,function(x) cor(x,cd8@meta.data$mass[cd8.24],method='spearman'))
cl3.corr.48 = apply(cd8@data[cc.idx,cd8.48],1,function(x) cor(x,cd8@meta.data$mass[cd8.48],method = 'spearman'))
boxplot(null.vec.24,cl3.corr.24,null.vec.48,cl3.corr.48, ylim = c(-0.7,0.7), outline=FALSE,col=c(rgb(0,0,0,0.5),rgb(0,0,1,1),rgb(0,0,0,0.5),rgb(1,0,0,1)), names = c("24h Null","24h Mass","48h Null","48 Mass"), ylab = "Pearson Coefficient",lwd=2,cex.lab = 1.5, cex.axis=1.5)