CENTRA: knowledge-based gene contextuality graphs reveal functional master regulators by centrality and fractality.

Deciphering gene function via context-aware approaches is limited by various means. Especially, static gene sets used in enrichment analyses and the lack of single-gene resolution restrain flexible association of genes with specific contexts. Here, we introduce CENTRA (Centrality-Based Exploration of Network Topologies from Regulatory Assemblies), a framework that models gene contextuality through topic-specific gene co-occurrence networks derived from curated gene sets and associated literature. Using latent dirichlet allocation on 12 045 abstracts linked to Molecular Signatures Database C2 gene sets, we uncover 27 biological topics and construct corresponding topic-specific networks reflecting distinct biological states, perturbation conditions, and disease-related regulatory programs. Graph-topological metrics, including centrality, local fractality, and perturbation sensitivity, were computed for each gene to capture structural relevance within these topic-specific networks. We show that topological profiles distinguish well-characterized regulators, identify emerging functional candidates, and reveal context-specific roles. Our framework prioritizes understudied genes by assessing the robustness of their topological signatures across topic-specific networks. To support exploration of these results, we developed a publicly accessible interactive browser, CENTRA, enabling dynamic navigation of networks and functional annotations. CENTRA provides an interpretable, scalable framework for investigating context-dependent gene function and hypothesis generation, offering a novel entry point beyond traditional enrichment approaches.