Alzheimer's disease biological domain sub-stratification enhances the precision of functional analyses

The Target Enablement to Accelerate Therapy Development for AD (TREAT-AD) bioinformatics pipeline employs a rank-and-organize strategy. Disease-associated genes drive enrichment of large AD-linked endophenotypes. However, these biological areas were too large to promote hypothesis development or target identification. Here we delineate subdomains that map to, and enrich, specific biological processes.

To refine the biodomains into more focused areas, we built κ networks out of the Gene Ontology terms in the biodomain and employ shared gene annotation between terms to determine edge weights. κ-value filtration enabled us to identify data-driven subdomains, which we employed in an analysis of TREAT-AD harmonized datasets.

The subdomain enrichment highlights core areas of biological impairment within the biodomain space and facilitates a deeper interpretation of large-scale multiomic datasets.

The subdomain mapping of AD-risk-associated processes may facilitate an open-source, open-science shareable resource for the comparison of large datasets for the formulation of future hypotheses and identification of therapeutic targets.