Roscoff 2018 / Single cell transcriptomics / Schlitzer et al.
Setup
To begin with, we will load a bunch of libraries that we will use for the analysis. Note the use of reticulate as we will be ussing functions from the sklearn library
library(readr)
library(magrittr)
library(ggplot2)
library(ElPiGraph.R)
library(reshape)
library(Rtsne)
library(igraph)##
## Attaching package: 'igraph'## The following objects are masked from 'package:stats':
##
## decompose, spectrum## The following object is masked from 'package:base':
##
## unionlibrary(rgl)
library(DDRTree)## Loading required package: irlba## Loading required package: Matrix##
## Attaching package: 'Matrix'## The following object is masked from 'package:reshape':
##
## expandlibrary(irlba)
library(reticulate)
sk <- import("sklearn.manifold")Now we will load the data available from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE60781. For this example we will use both the raw count and the RPKM data.
Let us begin by loading the raw count and formatting them.
BaseDir <- "~/Git/Roscoff/Datasets/Schlitzer/"
AllCells_RAW <- dir(path = paste0(BaseDir, "RAW"), full.names = TRUE)
AllCells_RAW <- AllCells_RAW[grep("readcount", AllCells_RAW, ignore.case = TRUE)]
CellRC <- list()
for(i in 1:length(AllCells_RAW)){
cat(i, " ")
CellRC[[i]] <- read.delim(file = AllCells_RAW[i], header = FALSE)
if(i != 1){
if(any(CellRC[[1]]$V1 != CellRC[[i]]$V1)){
print(paste("incpmpatible gene names for cell", i))
}
}
}## 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 251each element of the list is a 2D tibble. We begin by checking that all the files have the same genes.
all(
sapply(CellRC[-1], function(x){
all(x$V1 == CellRC[[1]]$V1)
})
)## [1] TRUESince this is indeed the case, we can combine the gene expression into a single gene count matrix and removed the original file. We also extract and save extra infor contained in the data.
ExpMat_RAW <- lapply(CellRC, function(x){
x$V2
})
ExpMat_RAW <- do.call(rbind, ExpMat_RAW)
colnames(ExpMat_RAW) <- CellRC[[1]]$V1
rownames(ExpMat_RAW) <- sapply(strsplit(AllCells_RAW, "_"), "[[", 2)
rm(CellRC)
gc()## used (Mb) gc trigger (Mb) max used (Mb)
## Ncells 1391872 74.4 2164898 115.7 1770749 94.6
## Vcells 5293970 40.4 24949369 190.4 25920948 197.8InfoCol <- (ncol(ExpMat_RAW)-4):ncol(ExpMat_RAW)
ExtraInfo <- ExpMat_RAW[,InfoCol]
ExpMat_RAW <- ExpMat_RAW[, -InfoCol]Now, we load the RPKM data
RPK_Expression <- read_delim(file = "~/Git/Roscoff/Datasets/Schlitzer/GSE60781_RPKM.txt.gz", col_names = TRUE, delim = "\t")## Parsed with column specification:
## cols(
## .default = col_double(),
## geneName = col_character()
## )## See spec(...) for full column specifications.Genes <- unlist(RPK_Expression$geneName)
RPK_Expression <- data.matrix(RPK_Expression[,-1])
rownames(RPK_Expression) <- Genes
RPK_Expression <- t(RPK_Expression)Just to make sure, we check that we have the same information
all(rownames(RPK_Expression) %in% rownames(ExpMat_RAW))## [1] TRUEall(rownames(ExpMat_RAW) %in% rownames(RPK_Expression))## [1] TRUEall(colnames(RPK_Expression) %in% colnames(ExpMat_RAW))## [1] TRUEall(colnames(ExpMat_RAW) %in% colnames(RPK_Expression))## [1] FALSEMissingGenes <- which(!(colnames(ExpMat_RAW) %in% colnames(RPK_Expression)))
head(ExpMat_RAW[, MissingGenes])## Aasdh Ablim2 Adam22 Add1 Bcl7b Cacnb2 Cd46 Clcn3 Cmah Dapk2 Ddo
## RMD072 0 0 0 0 107 0 0 0 0 0 0
## RMD073 0 0 0 460 1 0 0 0 0 0 0
## RMD074 0 0 0 0 0 0 0 8 0 0 0
## RMD075 0 0 0 0 0 0 0 0 0 0 0
## RMD076 10 0 0 3 2 0 0 0 0 0 0
## RMD077 0 0 0 0 0 0 0 0 0 0 0
## E2f3 Elf2 Fuk Gpx4 Grk6 H2-gs10 H2-t9 Hdac7 Htra2 Il15ra Kcnmb2
## RMD072 17 0 0 33 0 0 0 0 0 0 0
## RMD073 0 0 0 5 1 0 0 0 0 0 0
## RMD074 0 0 0 5 0 0 0 0 0 0 0
## RMD075 40 0 0 0 160 0 0 0 0 0 0
## RMD076 0 0 0 38 107 0 0 0 0 0 0
## RMD077 0 0 0 3 0 0 0 0 0 0 0
## Kif1b Lpcat4 Man2b2 Mat2b Mrpl13 Mrpl33 Mrpl4 Mrps11 Mrps17 Naaa
## RMD072 192 0 14 37 0 18 119 0 3 0
## RMD073 279 0 217 139 0 116 4 0 119 250
## RMD074 0 0 5 1 0 154 289 0 1 0
## RMD075 0 0 0 0 0 230 1 0 42 0
## RMD076 0 0 0 0 0 56 97 0 84 0
## RMD077 0 0 18 5 0 75 83 0 56 0
## Nox1 Nrcam Nrk Oscp1 Pink1 Pkd1l3 Plrg1 Rab22a Rab2a Rab2b Rab40c
## RMD072 0 0 0 0 0 0 0 0 0 0 0
## RMD073 0 0 0 0 2 0 0 0 65 0 0
## RMD074 0 0 0 0 0 0 0 0 517 0 0
## RMD075 0 0 0 0 0 0 0 0 6 0 0
## RMD076 0 0 0 0 0 0 0 0 0 0 0
## RMD077 0 0 0 0 0 0 0 0 5 0 0
## Rab4b Rab8b Rgs7 Sash1 Sh3d19 Sik2 Smc4 Sp1 Sprr2h Srpk1 Tcra Tcrb
## RMD072 5 0 0 0 0 0 567 20 0 2 0 0
## RMD073 22 0 0 0 0 0 647 0 0 0 0 0
## RMD074 33 0 0 0 0 0 22 94 0 0 0 0
## RMD075 0 1 0 0 0 0 1046 396 0 0 0 0
## RMD076 0 0 0 85 0 0 3 1 0 143 0 0
## RMD077 0 5 0 0 0 0 1128 919 0 22 0 0
## Tmeff2 Txnrd2 Ush2a Vav2 Yaf2 a2ld1 abca5 abca8a aprt art3 cacna1d
## RMD072 0 0 0 0 0 0 1 0 0 0 0
## RMD073 0 0 0 0 0 0 0 0 0 0 0
## RMD074 0 0 0 0 0 0 0 0 0 0 0
## RMD075 0 0 0 0 0 0 0 0 0 0 0
## RMD076 0 0 0 0 0 0 0 0 0 0 0
## RMD077 0 0 0 0 0 0 0 0 0 0 0
## cdc6 cml3 helB limd1 msi2 nek2 notch1 oasl1 piwiL4 rage scmh1
## RMD072 1 0 0 0 0 0 0 0 0 0 0
## RMD073 0 0 0 0 0 0 0 0 0 0 0
## RMD074 0 0 0 0 0 0 0 0 0 0 0
## RMD075 0 0 0 0 0 0 0 0 0 0 0
## RMD076 0 0 0 0 0 0 0 0 0 0 0
## RMD077 0 0 0 0 0 0 0 0 0 0 0
## slc18a2 slc26a6 slc43a2 smoc2 tubg1 wdr4 wiz
## RMD072 0 0 0 0 0 0 0
## RMD073 0 0 0 0 0 0 0
## RMD074 0 0 0 0 0 0 0
## RMD075 0 0 0 0 0 0 0
## RMD076 0 0 0 0 0 0 0
## RMD077 0 0 0 0 0 0 0Apparently not … lets continue anyway. The population asssigned to the cells can be obtained by the article. They are sequential, but there are some missing info.
CellID <- as.integer(sapply(strsplit(rownames(ExpMat_RAW), "RMD"), "[[", 2))
names(CellID)[CellID <= 167] <- "BM_CD115+_CDP"
names(CellID)[CellID >= 168 & CellID <= 263] <- "BM_PreDC"
names(CellID)[CellID >= 264] <- "BM_MDP"
names(CellID)[CellID == 78] <- "N/A"
names(CellID)[CellID == 89] <- "N/A"
names(CellID)[CellID == 100] <- "N/A"
names(CellID)[CellID == 121] <- "N/A"
names(CellID)[CellID == 137] <- "N/A"
names(CellID)[CellID == 158] <- "N/A"
names(CellID)[CellID == 160] <- "N/A"
names(CellID)[CellID == 166] <- "N/A"
names(CellID)[CellID == 186] <- "N/A"
names(CellID)[CellID == 203] <- "N/A"
names(CellID)[CellID == 208] <- "N/A"
names(CellID)[CellID == 220] <- "N/A"Feature selection
To simplify the analysis and to reduce noise, we will only select overdispersed genes. There are vaious approch to do this. Overall, we will apply the same as STREAM. To begin with we will filter genes expressed in less that 1% of the cells.
Thr <- round(.01*nrow(ExpMat_RAW))
table(colSums(ExpMat_RAW > 0) > Thr)##
## FALSE TRUE
## 17245 12618ToKeep <- colnames(ExpMat_RAW)[colSums(ExpMat_RAW > 0) > Thr]
ExpMat_RAW_Fil <- log2(ExpMat_RAW[, ToKeep] + 8)
RPK_Expression_Fil <- log2(RPK_Expression[, colnames(RPK_Expression) %in% ToKeep] + 8)and then select only overdispersed genes
X <- apply(ExpMat_RAW_Fil, 2, mean)
Y <- apply(ExpMat_RAW_Fil, 2, sd)
XSeq <- seq(from = 0, to = max(X), by=.1)
plot(X, Y, xlab = "mean gene expression", ylab = "sd of gene expression")
LOE <- loess(Y ~ X, span = .05)
points(XSeq, predict(LOE, newdata = data.frame(X = XSeq)), type = 'l', col = 'red')
Cols <- ifelse(predict(LOE) < Y, "red", "blue")
OEGenes <- colnames(ExpMat_RAW_Fil)[Y > predict(LOE)]
plot(X, Y, xlab = "mean gene expression", ylab = "sd of gene expression", col = adjustcolor(Cols, alpha.f = .4), pch = 20)
ExpMat_RAW_Fil_OE <- ExpMat_RAW_Fil[, OEGenes]We will repeat the same approach for RPKM data
X <- apply(RPK_Expression_Fil, 2, mean)
Y <- apply(RPK_Expression_Fil, 2, sd)
XSeq <- seq(from = 0, to = max(X), by=.1)
plot(X, Y, xlab = "mean gene expression", ylab = "sd of gene expression")
LOE <- loess(Y ~ X, span = .05)
points(XSeq, predict(LOE, newdata = data.frame(X = XSeq)), type = 'l', col = 'red')
Cols <- ifelse(predict(LOE) < Y, "red", "blue")
OEGenes_RPK <- colnames(RPK_Expression_Fil)[Y > predict(LOE)]
plot(X, Y, xlab = "mean gene expression", ylab = "sd of gene expression", col = adjustcolor(Cols, alpha.f = .4), pch = 20)
RPK_Expression_Fil_OE <- RPK_Expression_Fil[, OEGenes_RPK]We can also compare the selected overdispersed genes
table(OEGenes_RPK %in% OEGenes)##
## FALSE TRUE
## 1577 4709table(OEGenes %in% OEGenes_RPK)##
## FALSE TRUE
## 2462 4709Look at data
As a first step we check PCA
PCA <- prcomp_irlba(ExpMat_RAW_Fil_OE, n = 2, retx = TRUE)
PlotPCA <- data.frame(PCA$x, type = names(CellID))
ggplot(PlotPCA, aes(x = PC1, y = PC2, col = type)) + geom_point(size = 1, alpha=.7) + labs(title = "PCA of Raw data")
tSNE <- Rtsne(ExpMat_RAW_Fil_OE, initial_dims = 50, perplexity = 10)
PlotPCA <- data.frame(D1 = tSNE$Y[,1], D2 = tSNE$Y[,2], type = names(CellID))
ggplot(PlotPCA, aes(x = D1, y = D2, col = type)) + geom_point(size = 1, alpha=.7) + labs(title = "tSNE of Raw data")
tictoc::tic()
MLLE.Model <- sk$LocallyLinearEmbedding(n_neighbors=as.integer(ceiling(nrow(ExpMat_RAW_Fil_OE)*.2)),
n_components = as.integer(3),
method = 'modified',
eigen_solver = 'dense',
n_jobs = as.integer(1),
random_state = as.integer(10),
neighbors_algorithm = 'kd_tree')
MLLE.Fit <- MLLE.Model$fit(ExpMat_RAW_Fil_OE)
MLLE.Emb <- MLLE.Fit$embedding_
tictoc::toc()## 2.765 sec elapsedPlotPCA <- data.frame(D1 = MLLE.Emb[,1], D2 = MLLE.Emb[,2], type = names(CellID))
ggplot(PlotPCA, aes(x = D1, y = D2, col = type)) + geom_point(size = 1, alpha=.7) + labs(title = "MLLE of Raw data")
Pseudotime (MLLE)
boxplot(dist(MLLE.Emb))
Curve <- computeElasticPrincipalCurve(X = MLLE.Emb, NumNodes = 10, Do_PCA = FALSE, TrimmingRadius = .15, ProbPoint = .9, nReps = 100,
n.cores = 4, ClusType = "FORK", ParallelRep = TRUE)## [1] "Creating a sock cluster with 4 nodes"
## [1] "Using parallel sampling analysis. Limited output available"
## [1] "Generating the initial configurations"
## [1] "Exporting initial configuration parameters"
## [1] "Exporting initial configuration parameters"
## [1] "Graphical output will be suppressed"
## [1] "Exporting parameters"
## [1] "Analysis is running ... no output will be shown"
## [1] "Sit back and relax, this may take a long time ..."
## 3.582 sec elapsed
## [1] "Constructing average tree"
## [1] "Creating a chain in the 1st PC with 2 nodes"
## [1] "Using a user supplied cluster. Updating the value of X"
## [1] "Computing EPG with 10 nodes on 1000 points and 3 dimensions"
## [1] "Using a user supplied cluster. It must contains the data points in a matrix X"
## [1] "Exporting the additional variables to the cluster"
## Nodes = 2 3 4 5 6 7 8 9
## BARCODE ENERGY NNODES NEDGES NRIBS NSTARS NRAYS NRAYS2 MSE MSEP FVE FVEP UE UR URN URN2 URSD
## 0||10 0.00108 10 9 8 0 0 0 0.0006209 0.0005494 0.9234 0.9323 0.0002393 0.0002197 0.002197 0.02197 0
## 1.123 sec elapsed
## [[1]]
PlotPG(MLLE.Emb, Curve[[101]], BootPG = Curve[1:100], GroupsLab = names(CellID), DimToPlot = 1:3, NodeLabels = 1:30, LabMult = 4)## [[1]]
##
## [[2]]
##
## [[3]]
Extended <- ExtendLeaves(X = MLLE.Emb, TargetPG = Curve[[101]], ControlPar = .9, TrimmingRadius = .1) ## [[1]]
##
## [[1]]
PlotPG(MLLE.Emb, Extended, GroupsLab = names(CellID), DimToPlot = 1:3,
NodeLabels = 1:nrow(Extended$NodePositions), LabMult = 4)## [[1]]
##
## [[2]]
##
## [[3]]
Net <- ConstructGraph(PrintGraph = Extended)
Part <- PartitionData(X = MLLE.Emb, NodePositions = Extended$NodePositions)
PrjStruct <- project_point_onto_graph(X = MLLE.Emb,
NodePositions = Extended$NodePositions,
Edges = Extended$Edges$Edges,
Partition = Part$Partition)
Paths <- GetSubGraph(Net, "end2end")
Paths_Norm <- lapply(Paths, function(x){
if(x[1] != "12"){
rev(x)
} else {
x
}
})PT_MLLE <- getPseudotime(ProjStruct = PrjStruct, NodeSeq = names(Paths_Norm$Path_1))
hist(PT_MLLE$Pt)
SmoothGenes <- CompareOnBranches(X = ExpMat_RAW_Fil_OE, Paths = Paths_Norm, TargetPG = Curve[[1]],
GroupsLab = names(CellID), n.cores = 4, ClusType = "FORK",
Partition = Part$Partition, PrjStr = PrjStruct, Span = .2, Mode = "MI", ReturnGenes = TRUE)## [1] "Feature selection by mutual information"boxplot(SmoothGenes, ylab = "Mutual information")
CompareOnBranches(X = ExpMat_RAW_Fil_OE, Paths = Paths_Norm, TargetPG = Curve[[1]], GroupsLab = names(CellID),
Partition = Part$Partition, PrjStr = PrjStruct, Span = .3, Mode = "Var",
Features = names(SmoothGenes)[SmoothGenes > quantile(SmoothGenes, .98)])## [1] "16 pages will be plotted"## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
##
## [[5]]
##
## [[6]]
##
## [[7]]
##
## [[8]]
##
## [[9]]
##
## [[10]]
##
## [[11]]