Genedex: Contextualizing gene signatures across Alzheimer's disease pathology, cell types, and brain regions

Abstract

Systematic prioritization of candidate Alzheimer's disease (AD) genes and their target functions for therapeutic intervention requires a consistent understanding of their brain location, cell-type-specific expression, relationships with pathology, and direction of change. Defining the role of a gene in AD requires interrogation of multiple publications, each providing a different context. Often, it is also unclear whether a gene is up- or downregulated across these contexts. However, the narrow diversity of contexts in which they have been assayed and described limits interpretation. We integrated 1,562 gene signatures from 34 studies, annotated for pathological process, cell type, brain region, and direction of change, and showed that Genedex recovers established AD biology. Genedex addresses cellular and regional unexplained directional contradictions from 10,657 discordant genes in the published AD literature. We investigated the role of heat-shock chaperones and neurotrophic signaling in cognitive resilience using Genedex. Heat-shock chaperone genes are broadly upregulated with tau burden; but, in the presence of amyloid pathology, their expression diverges, showing upregulation in excitatory neurons and downregulation in glia and during A{beta} clearance. LINGO1, a regulator of BDNF, is upregulated with neuritic plaque load, tau burden, and selective vulnerability in cortical regions. Notably, in tau-affected RORB excitatory neurons, LINGO1 expression is inversely correlated with neurotrophic genes BDNF and NTRK2. A prioritized interneuronal resilience signature shows correlated pathway co-expression with somatostatin and parvalbumin interneuronal signatures. Vulnerability-associated signatures separate by cellular context: neuronal programs are positively correlated with vulnerable interneurons, while glial and vascular cells are predominantly anti-correlated. By systematically mapping genes across functional, cellular, spatial, and directional axes, Genedex provides a platform for discovery of contextualized targets and mechanistic hypotheses. Genedex is freely available at hidelab.github.io/genedex-app.