SC-02-20

SC-02-20.R

# Install Packages
install.packages(c('devtools','ggplot2'))
BiocManager::install(c('fgsea', 'MAST', 'ensembldb', 'GenomeInfoDb'))
devtools::install_github('cailab-tamu/scTypeGSEA')

# Loading libraries
library(Seurat)
library(MAST)
library(ggplot2)
library(scTypeGSEA)

# Moving to downloads
setwd("../Downloads/")

# Getting the input files from 10X Genomics
inputFile <- "http://cf.10xgenomics.com/samples/cell-exp/3.0.0/pbmc_1k_protein_v3/pbmc_1k_protein_v3_filtered_feature_bc_matrix.tar.gz"
download.file(inputFile, "PBMC.tar.gz")
untar("PBMC.tar.gz")

#Input Folder
inputFolder <- "filtered_feature_bc_matrix/"

# Loading Seurat
library(Seurat)

# Reading the dataset
PBMC <- Read10X(inputFolder)
PBMC <- PBMC[[1]]

# QC
mtReads <- PBMC[grepl("MT-", rownames(PBMC)),]
mtRate <- (apply(mtReads,2,sum)/apply(PBMC,2,sum))
plot(mtRate, pch=16)
abline(h = 0.1, col = "red", lty = 2)

# Filtering
PBMC <- PBMC[apply(PBMC,1,sum) != 0,]
PBMC <- PBMC[,mtRate < 0.1]

# Creating a Seurat Object
PBMC <- CreateSeuratObject(PBMC)

# Normalizing the dataset
PBMC <- NormalizeData(PBMC)
PBMC <- ScaleData(PBMC)

# Accessing to the raw data
PBMC@assays$RNA@counts

# Accessing to the normalized data
PBMC@assays$RNA@data

# Dimentionality Reduction
PBMC <- FindVariableFeatures(PBMC)
PBMC <- RunPCA(PBMC)
PCAPlot(PBMC)
PBMC <- RunUMAP(PBMC, dims = 1:50)
UMAPPlot(PBMC)
PBMC <- RunTSNE(PBMC, perplexity = 100)
TSNEPlot(PBMC)

# Finding the clusters
PBMC <- FindNeighbors(PBMC, do.plot = TRUE)
PBMC <- FindClusters(PBMC,resolution = 0.1)
UMAPPlot(PBMC)
TSNEPlot(PBMC)
PCAPlot(PBMC)

## Characterizing the clusters 
# Using Wilcoxon 
markerGenesW <- FindAllMarkers(PBMC, test.use = 'wilcox')

#Plotting
top10 <- lapply(unique(Idents(PBMC)), function(X){
  Xm <- markerGenesW[markerGenesW$cluster %in% X,]
  Xm <- Xm[order(Xm$avg_logFC, decreasing = TRUE),]
  Xm[1:10,7]
})
DoHeatmap(PBMC, features = unlist(top10), angle = 0, hjust = 2)
DotPlot(PBMC, features = unlist(top10)) + theme(axis.text.x = element_text(angle = 90, hjust = 1))

# Using MAST
markerGenesM <- FindAllMarkers(PBMC, test.use = 'MAST')
top10 <- lapply(unique(Idents(PBMC)), function(X){
  Xm <- markerGenesM[markerGenesM$cluster %in% X,]
  Xm <- Xm[order(Xm$avg_logFC, decreasing = TRUE),]
  Xm[1:10,7]
})
DoHeatmap(PBMC, features = unlist(top10), angle = 0, hjust = 2)
DotPlot(PBMC, features = unlist(top10)) + theme(axis.text.x = element_text(angle = 90, hjust = 1))

# scTypeGSEA
library(scTypeGSEA)
A <- scTypeGSEA::assignCellType(PBMC, test.use = 'MAST')
B <- scTypeGSEA::labelCelltype(PBMC, A$cluster_celltype)
PBMC <- B$obj
UMAPPlot(PBMC)
TSNEPlot(PBMC)

DoHeatmap(PBMC, features = unlist(top10), angle = 0, hjust = -10)
DotPlot(PBMC, features = unlist(top10)) + theme(axis.text.x = element_text(angle = 90, hjust = 1))