when is it appropriate to do a dotplot with your single-cell data?
I've seen people plot "by hand" genes that come up as markers from the clusters, but I was actually thinking about doing one with the most variable features, only, I don't know if that makes sense. I want to see what are the cells in each of my clusters doing.
FindMarkers or FindAllMarkers will produce most significant features of one cluster among a defined cluster or all the other clusters. You can assign top10-20 features to your dot plot for your purpose. "by hand" is to manually curate cluster annotation. Both are fine to proceed.
But in general, top features in one dataset may be totally different from another one. It might be due to technical limitation. In such case, you may find it is difficult to determine the cell type. Therefore, people commonly use "by hand" markers to manually annotate their clusters, which also require tremendous background knowledge.
There are many papers and protocols for cluster/cell annotation. Different labs use different methods. There is no standard method. It's up to you to decide which one you should choose.
