Disentangling Context-Specific Alzheimer's Disease Progression Signatures through an Explainable AI Framework

Alzheimer's disease (AD) progression is characterized by stage-dependent changes in gene expression, which can be obscured by overall cellular and biological fluctuations. Researchers have now developed ADVISE, an interpretable deep generative framework aimed at separating AD-specific transcriptional signals from background variations. This explainable AI approach emphasizes the genes and pathways responsible for disease-specific signatures.

By applying ADVISE to extensive single-nucleus RNA sequencing data from human prefrontal cortex samples, the study reveals both shared and unique AD progression patterns across the entire brain, differentiating between males and females, and varying by cell type. These analyses uncover converging molecular pathways linked to AD, such as those involved in synaptic transmission, HSF1-associated stress responses, and axon guidance via SLIT/ROBO proteins.

Remarkably, the framework singles out ROBO2 as a significant candidate for progression in excitatory neurons.

To further validate these findings, the research team utilized CRISPR activation on human induced pluripotent stem cell-derived excitatory neurons. The results show that activating ROBO2 decreases the ratio of phosphorylated Tau to total Tau, a key biomarker of AD pathology. Therefore, ADVISE not only uncovers context-dependent molecular signatures of AD progression but also paves the way for experimental validation of potential therapeutic targets for Alzheimer's disease.