Fig. 1: Schematic of RSS-NET.
From: Modeling regulatory network topology improves genome-wide analyses of complex human traits
a Decomposition of the total effect of a common SNP on a complex trait through multiple nearby and distal genes. b Gene regulatory network defined as a weighted and directed bipartite graph linking TFs to TGs. c RSS-NET exploits the topology of a TF-TG network to decompose the total genetic effect into cis and trans-regulatory components. Both the SNP-gene (cjg) and TF-TG (vgt) weights in this decomposition are assumed known and are specified by existing omics data (Methods). In addition to TF-TG networks, RSS-NET also requires d GWAS summary statistics and e ancestry-matching LD estimates as input. f Bayesian hierarchical model underlying RSS-NET. An in-depth description is provided in Methods. g Given a network, RSS-NET produces a Bayes factor comparing the baseline (M0) and enrichment (M1) models to summarize the evidence for network enrichment. h RSS-NET prioritizes loci within an enriched network by computing P1, the posterior probability that at least one SNP j in a locus is trait-associated (βj ≠ 0). Differences between P1 under M0 and M1 reflect the influence of a regulatory network on genetic associations, highlighting previously undescribed trait-associated genes.
