I would like to know what the experts feel about defining single-cell clusters using up-regulated versus down-regulated marker genes. There is this general view that down-regulated marker genes are ambiguous (and justifiably so), and hence clusters are much better defined by considering only the up-regulated marker genes compared to the rest of the cells in the data. While I understand the overall reasoning behind this, packages do offer the option of obtaining down-regulated marker genes as well. So, what are the caveats of including down-regulated genes in the list? If any, what are the advantages?
Would really appreciate any replies!