Network‐based Genetic Subtyping for Target Prioritization and Drug Repositioning: The Artificial Intelligence for Alzheimer’s Disease (AI4AD) Consortium

Abstract

AbstractBackgroundAlzheimer disease (AD) is a progressive neurodegenerative disease with complicated underlying disease mechanisms. The clinical and neuropathological heterogeneity of AD may be explained by specific disease mechanisms at a brain cell‐level. We postulate that the path to effective treatment of AD depends on identifying distinctive subtypes linking to specific disease mechanisms at brain cellular level.MethodWe developed a novel precision medicine framework using polygenic risk scores defined by co‐regulated cellular networks to stratify individuals into low and high‐risk groups (genetic subtypes) for brain cell‐type specific networks. We prioritized genes from the cell‐based networks using a graph‐based ranking (PageRank) algorithm. We investigated molecular signatures of the prioritized genes for the subtypes using autopsied brains from the Framingham Heart Study (FHS). We validated the signatures using isogenic APOE lines of human pluripotent stem cell (hiPSC)‐derived astrocytes. Existing drugs targeting for the genetic subtypes were identified using drug perturbation database and clustering algorithms using compound structures from the PubChem databaseResultWe identified six preserved and AD associated co‐expressed gene networks for astrocytes, oligodendrocytes, and oligodendrocyte progenitor cells. The top‐ranked astrocyte network was enriched with Parkinson’s disease, Alzheimer’s disease, mineral absorption, and ribosome pathways. The most important gene in the astrocyte network was the APOE gene, which its expression was increased in AD compared in control brains with elevated complement C4A level. We confirmed elevated expression of both APOE and C4A levels in APOE4 compared in APOE3 hiPSC‐derived astrocytes. We identified three drugs targeting the astrocyte subtype, all of them previously modulated expression of the prioritized genes in the drug perturbation database. Clustering algorithm identified multiple compounds targeting the prioritized genes in the astrocyte network with at least 80% similarity in structure.ConclusionThe newly established framework provided the potential implication in genetic subtype driven drug repositioning in AD. In future, we will develop novel graph neural network approaches to identify comprehensive genetic subtypes and drugs targeting the subtypes in collaboration with internal and external members from the AI4AD consortium.