raymondlouie
commented
1 year ago 1) Some of these issues may potentially be addressed by making it clear what our output is in the readme, and in the paper. 2) We can add an additional output, where we do a differential test of the clusters, and only show the markers we have identified (using Seurat for example). We can also propose a gating procedure using our markers for the clusters?.
It’s a bit unclear what the output of all the markers is referring to. E.g., when we run “all clusters” in Seurat I would expect a list of markers that are enriched in each of the clusters relative to all other clusters. In ClusterMarkers, we get one list of markers. What is the output? Is it the markers that are the most variable across all the clusters? Which clusters are they most enriched in? If users are wanting an output that is similar to Seurat (and you want them to use this package over Seurat), it may be worth including a similar output.