Cross-species comparison with ExpiMap

[pikapika505]

I am interested in using ExpiMap to identify shared and unique gene expression programs in scRNA-seq data from two species. All samples are from tumors, so I don't have any healthy or control samples. I previously attempted to use species1 as a reference and species2 as a query to detect differentially enriched GEPs, but this approach didn't yield the expected results. I have a few questions:

• Is it possible to identify differentially enriched GEPs between the reference and query datasets if I don't have any perturbations in datasets?
• Do the latent_directions output values indicate enrichment status, with 0 for neutral, -1 for downregulated, and 1 for upregulated?
• To project GEPs onto UMAP, I understand that I need to calculate cell embeddings using the get_latent function, but it doesn’t seem to accept "ref/query" as labels. Is there a workaround for this?
• If I want to examine gene weights in each term within the reference dataset, I can use the term_genes function. However, is there a way to determine which GPs are highly expressed in a specific dataset? Is there any scoring method for enrichment or importance of such GPs?

Thank you for your help!