As always, thank you for this amazing resource.
I have noticed that the most important “axes” of my data naturally align along UMAP1 and/or UMAP2 after clustering, which is super cool! I can sort of get at the genes that define these axes by picking clusters at different extremes. But, I’m curious if there is a way to directly access the weights of genes that comprise UMAP1 and UMAP2. Perhaps this is simply the weights from the PCA components?
Example, let’s say I have a single elongated (oval) cluster oriented long-ways up and down, I’d like to say that the top 10 genes that define the UMAP2 axis (up-down) are X1,X2…X10.