immunogenomics
Correct procedure for subclustering
Hi,
Thank you so much for this great tool!
Hope you can help me with this question. I am working with a scRNA-seq dataset in which I have been using SCTransformation as a normalization approach and harmony as a batch-correction tool. Now I have begun subclustering my dataset, however I am very uncertain about the correct approach for subclustering. After subsetting my data, I suppose that the best practice would be to re-integrate the subset, however I don't know if I should re-normalize the data as well?
My two proposed pibelines are:
susbetting, normalization og normal integration: P165_singlets <- subset(final_dp_seurat_Pat165, subset = DF.classification == 'Singlet') P165_singlets <- SCTransform(P165_singlets, vars.to.regress = c('percent.mt', 'nCount_RNA'), verbose = FALSE) P165_singlets <- RunPCA(P165_singlets, features = VariableFeatures(object = P165_singlets)) P165_singlets <- RunHarmony(P165_singlets, 'orig.ident', assay.use='SCT', max.iter.harmony = 20) P165_singlets <- FindNeighbors(P165_singlets, dims = 1:40) P165_singlets <- FindClusters(P165_singlets, resolution = 0.15) P165_singlets <- RunUMAP(P165_singlets, dims = 1:40, n.neighbors = 15, umap.method = 'umap-learn')
Subsetting, no-normalization og normal integration P165_singlets <- subset(final_dp_seurat_Pat165, subset = DF.classification == 'Singlet') P165_singlets <- RunPCA(P165_singlets, features = VariableFeatures(object = P165_singlets)) P165_singlets <- RunHarmony(P165_singlets, 'orig.ident', assay.use='SCT', max.iter.harmony = 20) P165_singlets <- FindNeighbors(P165_singlets, dims = 1:40) P165_singlets <- FindClusters(P165_singlets, resolution = 0.15) P165_singlets <- RunUMAP(P165_singlets, dims = 1:40, n.neighbors = 15, umap.method = 'umap-learn')
I think this issue has been widely discussed, but I haven't managed to find a concrete answer!
Best regards, Peter
