Hi everybody, I have a conceptual question, reading online, it seems that Seurat's integration with IntegrateLayers works only in low dimensional space, integrateData on the other hand modifyes the expression matrix.
It also seems that using any of these methods is not different methodoligically for any downstream analysis. I don't understand how this is possible, for example when doing an analysis like identifiyng the differentially expressed genes, or anything that works with the expression matrix.
As far as I can understand, it would not make sense to do work with it as the integration only happens on the low dimensional space. This would mean that I can only perform operations such as clustering, but nothing that works with the expression matrix as it is just like before integration. I believe to be wrong, but I don't understand how.
Thank you in advance,
