Hello all! I would like to preface this post by saying that I am not a bioinformatician and I need help with the clustering of a scRNA-seq PBMC dataset and generating a UMAP with different annotated cell types. I am running the the R script below and have annotated the cell-types with SingleR using the Monacco Immune Dataset to identify broad cell types (B cells, T cells, NK cells etc.). However in my UMAP all the cell types are intermixed and I would like them to form separate clusters.
# 1. Normalization -----
pbmc <- NormalizeData(pbmc)
# 2. Find variable features -----
pbmc <- FindVariableFeatures(pbmc, nfeatures = 5000)
# 3. Scale data -----
pbmc <- ScaleData(pbmc, vars.to.regress = "percent.mt")
# 4. Compute PCA -----
pbmc <- RunPCA(pbmc)
# 5. Determine dimentionality of the dataset (how many PCs to capture?) -----
ElbowPlot(pbmc, ndims = 50)
# 6. Compute neighbors and clusters -----
pbmc <- FindNeighbors(pbmc, dims = 1:20)
pbmc <- FindClusters(pbmc, resolution = 0.095, group.singletons = F, algorithm = 1)
# 7. Build Cluster tree -----
pbmc <- BuildClusterTree(pbmc, reorder = T, reorder.numeric = T)
# 8. Run UMAP -----
pbmc <- RunUMAP(pbmc, dims = 1:20, n.neighbors = 20)
# 9. Plot UMAP colored by SingleR labels -----
DimPlot(pbmc,
reduction = "umap",
raster = F,
label = T,
repel = T) | DimPlot(pbmc,
reduction = "umap",
raster = F,
repel = T,
group.by = "cell_type") + ggtitle("Cell type annotation")
Can someone please advice me what changes to make to the code to separate the clusters as it looks like there is B cell expression everywhere. I have tried playing around with the resolution and algorithm used in FIndClusters but so far nothing works for me.
Can you share the part where you ran SingleR on your data? You might also need to share some QC plots and details on the library prep/sequencing technology