Hi,
From my understanding, if you have two samples of single cell rna sequencing data, you should merge them to log normalize and scale. Then split them to find variable features and sctransform for eventual integration and downstream clustering. My question is, once you have identified initial clusters and are planning to sub cluster and have subsetted your cells that you want to subcluster, should you follow the same procedure of normalization, sctransform, and integration for both samples before subclusetring? Or since they had been initially normalized and integrated that was enough and now I can subset the cells from let’s say cluster 1 to go straight to subclustering? Appreciate the any help as I am trying to decide the best approach on how to create my workflow.
You don't need to integrate again. Setting the assay to "integrated" and re-clustering should work. Keep in mind that you cannot simply run
FindClusters, but you need to re-calculate PCA,uMAP and Neighbors first.Thank you for your reply and guidance. Just to double check, I wouldn’t have to normalize and scale the subsetted cells either when reclustering, just runPCA and UMAP before running FindClusters? Appreciate your help very much.