Abstract
Many common diseases are highly polygenic, modulated by a large number of genetic factors with small effects on susceptibility to disease. These small effects are difficult to map reliably in genetic association studies. To address this problem, researchers have developed methods that aggregate information over sets of related genes, such as biological pathways, to identify gene sets that are enriched for genetic variants associated with disease. However, these methods fail to answer a key question: which genes and genetic variants are associated with disease risk? We develop a method based on sparse multiple regression that simultaneously identifies enriched pathways, and prioritizes the variants within these pathways, to locate additional variants associated with disease susceptibility. A central feature of our approach is an estimate of the strength of enrichment, which yields a coherent way to prioritize variants in enriched pathways. We illustrate the benefits of our approach in a genome-wide association study of Crohn's disease with ~440,000 genetic variants genotyped for ~4700 study subjects. We obtain strong support for enrichment of IL-12, IL-23 and other cytokine signaling pathways. Furthermore, prioritizing variants in these enriched pathways yields support for additional disease-association variants, all of which have been independently reported in other case-control studies for Crohn's disease.
Introduction
Most common diseases, such as Crohn’s disease or diabetes, are polygenic, influenced by hundreds or thousands of genetic variants with individually small effects. Standard Genome-Wide Association Study (GWAS) methods typically test one variant at a time, often failing to reach the strict statistical thresholds required for significance. While pathway analysis has emerged as a way to group related genes to find stronger signals, these methods often act as a dead end—identifying an involved pathway without pinpointing the specific driving variants. This paper introduces a unified Bayesian framework that uses pathway-level information as an informative prior to improve the discovery of individual disease-associated variants.
Methodology
The authors move away from single-marker testing toward a sparse multiple regression model that considers the joint effects of all genetic variants simultaneously.
- Bayesian Variable Selection: The model calculates a Posterior Inclusion Probability (PIP)—the probability that a specific variant is truly associated with the disease.
- The Enrichment Parameter ($\theta$): This core innovation increases the prior probability of a variant being included if it belongs to a candidate pathway.
- Bayes Factors (BF): Thousands of pathways are ranked based on the evidence for their enrichment.
- Variational Approximation: To handle the massive computational burden of ~440,000 SNPs across ~3,000 pathways, the authors use a fast variational inference approach rather than traditional MCMC methods.
Results
Identification of Enriched Pathways
Applying this model to Crohn’s disease data from the Wellcome Trust Case Control Consortium (WTCCC), the analysis found overwhelming support for the enrichment of cytokine signaling pathways. Specifically, IL-12 and IL-23 mediated signaling events showed the strongest evidence for association.
Variant Prioritization and Discovery
By using enrichment as an informative prior, the model "up-weights" variants within relevant biological contexts. This allows for the discovery of associations that were not significant in the original analysis of the same data. Loci such as STAT3, the IBD5 locus, and MHC Class II genes showed significantly higher PIPs when the cytokine signaling pathway was included in the model compared to a null model with no enrichment.
Crucially, all of these newly prioritized associations have since been confirmed by independent, larger meta-analyses, validating the model's ability to extract more "truth" from smaller datasets. The authors estimate that approximately 47 independent risk factors for Crohn's disease may be hidden within the cytokine signaling pathway alone.
Discussion
This framework bridges the gap between broad pathway analysis and specific variant discovery. By treating biological knowledge as a prior rather than a post-hoc check, the model effectively navigates the polygenic landscape of disease. While the absolute Bayes Factors can be sensitive to prior specifications for effect sizes, the relative ranking of pathways remains robust. Future work may adapt this framework for rare variant analysis in whole-genome sequencing data, where single-marker tests are even less powerful.
Summary updated