'keys' in get_mapping_scores and GenePairFinder #161
Description
Hi Alec, thank you very much for this valuable tool! I've got a couple of perplexities on the 'keys' parameter when used inside the get_mapping_scores and GenePairFinder functions.
Basically, I am running SAMap on single-cell atlases of several distantly-related species, to uncover affinities among cell types. The majority of the datasets I am using has been already published in a previous work, and at the moment I am not willing to compute and annotate cell clusters de novo. Therefore, I am keeping the cell cluster annotation in the AnnData.obs object.
I run SAMAP using the leiden clustering method with default parameters , and everything went well. Then, I run the get_mapping_scores and GenePairFinder functions to get "fancier data" to plot. However, in these steps, I got the parameters 'keys' to point at the cluster annotation column in the AnnData.obs object. As expected, the obtained results are referring to the annotated cell clusters (connecting for example cell-cluster-1 from species-1 to cell-cluster-4 from species-2), rather than to the computed leiden cluster.
My question now is: are these results reliable? How does the algorithm handle the usage of a costum cluster annotation in the calculation of mapping scores and gene pairs, when it was previously instructed to use leiden clusters to build the stitched space? My doubt arose because of course I expect that there might not be a 1:1 relationship between the leiden clusters and the costum clusters, but rather a many-to-many. Maybe it is a silly one, but I just want to make sure that I am not mis-interpreting data or that results are somwhow biased.
Hope I made my point clear. Happy to share further data on this. And thnak you for the help!