scAGG produces cell-level embeddings of single-cell data, optimized for the classification of Alzheimer’s disease. We then correlate the 1st principal component with gene expression of all genes to find genes that are associated with the learned spectrum between AD and healthy controls. See the publication for exact methodology: https://doi.org/10.1016/j.csbj.2025.08.009.
Access: the cell-type specific gene rankings (.csv) are available in the BRAINSCAPES Zenodo, see: https://zenodo.org/records/17455417.
