scRNA cross-species integration using Seurat v3

Question

[maddhesiya13]

Hi,

I am trying to integrate zebrafish and mouse genome using Seurat v3. I am getting an error while running the anchoring step (FindIntegrationAnchors). I have the orthologous table between the two species obtained from biomart however I don't understand at what step should I upload it to reannotate the genes in order to get common features? Both species samples were processed separately following https://satijalab.org/seurat/archive/v3.2/pbmc3k_tutorial.html and for integration https://satijalab.org/seurat/archive/v3.2/integration.html I am not an expert on this perhaps I am missing something usual. Any suggestions would be highly appreciated.

Many thanks!

Answer 1

[ATpoint]

I am not a Seurat user but intuitively (unless there is a dedicated function for it) you probably have to first subset both datasets to only include genes that actually have homologs between each other, and then replace the original gene names with the ortholog names from the table so both datasets have the same gene names, because from what I know that is what is required so that the integration function can work. Otherwise the function has no way of telling which genes are actually "the same" so orthologs between the species.

[maddhesiya13]

Thanks for your suggestion, I actually tried that way, unfortunately, didn't work. You are right, for integration common features (genes) required, however, I realised the letter-case of genes differ between the species (Mouse genome (e.g. Olig2) and zebrafish (e.g. olig2)). How should I tackle this?