Detecting latent interaction effects when analyzing binary traits
Ziang Zhang,Jerald F. Lawless,Andrew D. Paterson ,Lei Sun
Abstract
In genome-wide association studies (GWAS), it is often desirable to test for interactions, such as gene–environment (G x E) or gene–gene (G x G) interactions, between single-nucleotide polymorphisms (SNPs, G’s) and environmental variables (E’s). However, directly accounting for interaction is often infeasible, because the interacting variable is latent or the computational burden is too large. For quantitative traits (Y) that are approximately normally distributed, it has been shown that indirect testing on GxE can be done by testing for heteroskedasticity of Y between genotypes.
Introduction
It is well known that the interaction (denoted as GxE) between single-nucleotide polymorphisms (SNPs; G’s) and environmental factors (E’s), or between SNPs (denoted as GxG), play an important role in shaping human complex traits (Y’s) [14]. A classic GxE example is the interaction effect between genetic variants in PAH and diet on the risk of phenylketonuria and its subsequent intellectual disability [11]. Examples of GxG have also been reported by [21].
Materials and method
Ethics Statement.
This research has been conducted using the UK Biobank Resource under Application Number 64875. The ethics approval of UK Biobank has been obtained from the North West Multi-centre Research Ethics Committee (MREC).
Results
We illustrated the usage of the proposed non-additive (indirect) test and its subsequent (2-df) joint test. We achieved this by a GWAS on UKB of the binary trait (self-reported) of high cholesterol (Data-Field 20002; Coding 1473) [1,24].
Discussion
Using heteroskedasticity to indirectly test for a latent interaction is well-established in the analysis of quantitative traits, and has led to many scientific insights over the human genome. However, none of the existing approaches of indirect testing could be applied when the trait of interest is binary. In this paper, we (i) derive, for the first time, an indirect test for binary traits, and in doing so, (ii) offer a practical interpretation for non-additive effects identified in binary trait GWAS.
Acknowledgments
Ziang Zhang was a trainee of the CANSSI-ONTARIO STAGE (Strategic Training for Advanced Genetic Epidemiology) training program at the University of Toronto.
Citation: Zhang Z, Lawless JF, Paterson AD, Sun L (2025) Detecting latent interaction effects when analyzing binary traits. PLoS Genet 21(8): e1011822. https://doi.org/10.1371/journal.pgen.1011822
Editor: Heather J. Cordell, Newcastle University, UNITED KINGDOM OF GREAT BRITAIN AND NORTHERN IRELAND
Received: December 16, 2024; Accepted: July 24, 2025; Published: August 22, 2025
Copyright: © 2025 Zhang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: This research has been conducted using the UK Biobank Resource under Application Number 64875. Data are available at https://www.ukbiobank.ac.uk with the permission of UK Biobank. The code used to replicate the analysis results in this paper can be accessed at github.com/AgueroZZ/GE_code_repo. The ethics approval of UK Biobank has been obtained from the North West Multi-centre Research Ethics Committee (MREC). The GWAS summary statistics described in Sect 5 are publicly available at https://zenodo.org/records/16279185.
Funding: This research was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC; url: https://www.nserc-crsng.gc.ca/index_eng.asp; Grant Number: RGPIN-04934 to LS), the Center for Addiction and Mental Health Discovery Fund Seed Funding (CAMH; url: https://www.camh.ca/en/science-and-research/discovery-fund/seed-funding-projects; to LS), the University of Toronto Data Sciences Institute Catalyst Grant (DSI; url: https://datasciences.utoronto.ca/; to LS), and the Canadian Statistical Sciences Institute (CANSSI-STAGE; url: https://stage.utoronto.ca/; to ZZ). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.'
Competing interests: The authors have declared that no competing interests exist.
