Steps involved in Subsetting clusture of interest and reclusturing

Method1

Method2

Subset Clustures

Step1 : Loading and plotting the analysed data

library(Seurat)
CombinedData <- readRDS("CombinedData_labeled.rds")
png(filename = "CombinedData.png")
DimPlot(CombinedData, reduction = "umap")
dev.off()

Step2 : Subsetting the clustures of interest

CombinedDataSubset<-subset(CombinedData, ident = c("CD14 Mono", "CD16 Mono", "pDCs", "cDCs"))

Check the Subset plot

png(filename = "CombinedDataSubset.png")
DimPlot(CombinedDataSubset, reduction = "umap")
dev.off()
saveRDS(CombinedDataSubset,file="CombinedDataSubset.Rds")

Method1

Step3 :Re-clusturing the subset

CombinedDataSubset <- ScaleData(CombinedDataSubset, verbose = FALSE)
CombinedDataSubset <- RunPCA(CombinedDataSubset, npcs = 30, verbose = FALSE)
CombinedDataSubset <- RunUMAP(CombinedDataSubset, reduction = "pca", dims = 1:20)
CombinedDataSubset <- RunTSNE(CombinedDataSubset, reduction = "pca", dims = 1:20)
CombinedDataSubset <- FindNeighbors(CombinedDataSubset, reduction = "pca", dims = 1:20)
CombinedDataSubset <- FindClusters(CombinedDataSubset, resolution = 1.0)#resolution was set to 1.0 because this is a large dataset

Step4 : Plot the new clustures based on the subset

png(filename = "CombinedData_Reclustured.png")
DimPlot(CombinedDataSubset, reduction = "umap")
dev.off()

Method2

Step3 : Change the defualt assay

DefaultAssay(CombinedDataSubset) <- "RNA"

Step4 : Split the Data for reanalysis based on Visits

For downstream analysis each sample/condition need minimum of 200 cells. else its better to remove those samples or merge with other samples

Subset_Cells.list <- SplitObject(CombinedDataSubset, split.by = "Visits")

Step5 : Rerun SCT

for (i in 1:length(Subset_Cells.list)) {
  Subset_Cells.list[[i]] <- PercentageFeatureSet(Subset_Cells.list[[i]], pattern = "^MT-", col.name = "percent.mt")
  Subset_Cells.list[[i]] <- SCTransform(Subset_Cells.list[[i]], vars.to.regress = "percent.mt", verbose = FALSE)
  #Subset_Cells.list[[i]] <- NormalizeData(Subset_Cells.list[[i]], verbose = FALSE)
  #Subset_Cells.list[[i]] <- FindVariableFeatures(Subset_Cells.list[[i]], selection.method = "vst", nfeatures = 2000,  verbose = FALSE)
}

Step6 : Finding anchors and standard processing

pag_combined.anchors <- FindIntegrationAnchors(object.list = Subset_Cells.list, dims = 1:6)
CombinedDataSubset.combined <- IntegrateData(anchorset = pag_combined.anchors, dims = 1:6)
CombinedDataSubset.combined <- ScaleData(CombinedDataSubset.combined, verbose = FALSE)
CombinedDataSubset.combined <- RunPCA(CombinedDataSubset.combined, npcs = 6, verbose = FALSE)
CombinedDataSubset.combined <- RunUMAP(CombinedDataSubset.combined, reduction = "pca", dims = 1:6)
CombinedDataSubset.combined <- RunTSNE(CombinedDataSubset.combined, reduction = "pca", dims = 1:6,check_duplicates = FALSE)
CombinedDataSubset.combined <- FindNeighbors(CombinedDataSubset.combined, reduction = "pca", dims = 1:6)
CombinedDataSubset.combined <- FindClusters(CombinedDataSubset.combined)

Step7: Plot the UMAP

png(filename = "CombinedData_Reclustured_BYVisit.png")
DimPlot(CombinedDataSubset.combined, reduction = "umap")
dev.off()