LSMM: A statistical approach to integrating functional annotations with genome-wide association studies

TL;DR

This paper introduces LSMM, a statistical model that integrates functional annotations with GWAS data to improve the identification of risk variants and provide biological insights into complex traits.

Contribution

The study presents a novel latent sparse mixed model (LSMM) with an efficient variational EM algorithm for scalable integration of functional annotations with GWAS data.

Findings

LSMM outperforms conventional methods in statistical power.

Application to real GWAS data reveals biologically relevant annotations.

LSMM enhances understanding of genetic architecture of complex traits.

Abstract

Thousands of risk variants underlying complex phenotypes (quantitative traits and diseases) have been identified in genome-wide association studies (GWAS). However, there are still two major challenges towards deepening our understanding of the genetic architectures of complex phenotypes. First, the majority of GWAS hits are in the non-coding region and their biological interpretation is still unclear. Second, accumulating evidence from GWAS suggests the polygenicity of complex traits, i.e., a complex trait is often affected by many variants with small or moderate effects, whereas a large proportion of risk variants with small effects remains unknown. The availability of functional annotation data enables us to address the above challenges. In this study, we propose a latent sparse mixed model (LSMM) to integrate functional annotations with GWAS data. Not only does it increase statistical power of the identification of risk variants, but also offers more biological insights by detecting relevant functional annotations. To allow LSMM scalable to millions of variants and hundreds of functional annotations, we developed an efficient variational expectation-maximization (EM) algorithm for model parameter estimation and statistical inference. We first conducted comprehensive simulation studies to evaluate the performance of LSMM. Then we applied it to analyze 30 GWAS of complex phenotypes integrated with 9 genic category annotations and 127 tissue-specific functional annotations from the Roadmap project. The results demonstrate that our method possesses more statistical power over conventional methods, and can help researchers achieve deeper understanding of genetic architecture of these complex phenotypes.