回溯亚群细分效果到初始分群

library(Seurat)
library(ggplot2)

# Remove variables
rm(list = ls())

# Set workdir
workdir <- 'F:/R_Language/R_Practice/scRNA_Seq_column'
setwd(workdir)

# Load seurat object variable
data.filt <- readRDS(file = "data/Raw_data/Cluster_all_again.Rdata")
print(table(Idents(data.filt)))

## 
##   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16 
## 406 399 352 340 332 283 247 208 184 181 159  81  56  47  41  37

DT::datatable(data.filt@meta.data)

sub_cells <- subset(data.filt, cluster == 1)
DT::datatable(sub_cells@meta.data)

# Determine the K-nearest neighbor graph
sub_cells <- FindNeighbors(sub_cells, reduction="pca", dims = 1:20, verbose = F)
# Determine the clusters for various resolutions
sub_cells <- FindClusters(sub_cells, resolution = 0.5, verbose = F)

# t-SNE and Clustering
sub_cells <- RunTSNE(sub_cells, reduction = "pca", dims = 1:20, verbose = F) # check_duplicates = FALSE, Remove duplicates before running TSNE
sub_cells <- RunUMAP(sub_cells, reduction = "pca", dims = 1:20, verbose = F)

## Rename sub cells
metadata <- sub_cells@meta.data
metadata$cluster <- Idents(sub_cells)
metadata$sub_cluster <- paste0('M', metadata$cluster)
sub_cells@meta.data <- metadata
table(sub_cells$sub_cluster)

## 
##  M0  M1  M2 
## 170 151  85

## Mapping to old clusters
metadata <- data.filt@meta.data
metadata$cell_type <- NA
metadata$cell_type <- sub_cells@meta.data[match(rownames(data.filt@meta.data), rownames(sub_cells@meta.data)), 'sub_cluster']
print(table(metadata$cell_type))

## 
##  M0  M1  M2 
## 170 151  85

DT::datatable(metadata)

celltype_names <- NULL
for(i in 1:dim(metadata)[1]){
  # print(i)
  sub_data <- metadata[i,]
  if(is.na(sub_data$cell_type)){
    # print('Change value')
    sub_data$cell_type <- sub_data$cluster
    celltype_names <- c(celltype_names, sub_data$cell_type)
  }else{
    celltype_names <- c(celltype_names, sub_data$cell_type)
  }
}

print(table(celltype_names))

## celltype_names
##  10  11  12  13  14  15  16   2   3   4   5   6   7   8   9  M0  M1  M2 
## 181 159  81  56  47  41  37 399 352 340 332 283 247 208 184 170 151  85

metadata$cell_type <- celltype_names
metadata$cell_type <- factor(metadata$cell_type)
metadata$cell_type <- factor(metadata$cell_type, levels = c(names(table(sub_cells$sub_cluster)), 2:nlevels(Idents(data.filt))))
print(table(metadata$cell_type))

## 
##  M0  M1  M2   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16 
## 170 151  85 399 352 340 332 283 247 208 184 181 159  81  56  47  41  37

data.filt@meta.data <- metadata 

DimPlot(data.filt, reduction='tsne', label=T, label.size=5, pt.size=1, group.by='cluster') + ggtitle('original celltype')

DimPlot(data.filt, reduction='tsne', label=T, label.size=5, pt.size=1, group.by='cell_type') + ggtitle('new celltype')