PIGEON: a statistical framework for estimating gene-environment interaction for polygenic traits

GWAS
gene-environment interaction
gxe
ld score regression
polygenic traits
statistical methods
summary statistics
variance component
  • PIGEON Framework: The paper introduces PIGEON, a unified statistical framework that uses a variance component analytical approach to quantify polygenic Gene-Environment Interaction (GxE).
  • Input and Scalability: PIGEON is highly scalable as its estimation procedure requires only GWAS and GWIS summary statistics (not individual-level data).
  • Key Objectives: The framework provides rigorous methods for both detecting the presence of GxE (by estimating GxE variance) and interpreting its mechanism (by estimating Oracle PGSxE), demonstrating its utility across multiple empirical settings including gene-by-sex and gene-by-education studies.
Published

23 January 2026

PubMed: 40410536 DOI: 10.1038/s41562-025-02202-9 Overview generated by: Gemini 2.5 Flash, 28/11/2025

Key Findings and Motivation

Understanding Gene-Environment Interaction (GxE) is crucial for deciphering the genetic architecture of human complex traits. The authors introduce the PIGEON (PolygenIc Gene-Environment interactiON) framework to address the challenges in scalability and interpretability faced by current GxE methods. PIGEON provides a unified and statistically grounded approach for quantifying polygenic GxE effects, requiring only summary statistics data as input.

Study Design and Methods

PIGEON Framework

PIGEON is a unified statistical framework designed to model polygenic GxE effects for complex traits using a variance component analytical approach. It allows researchers to define the parameters of interest and systematically compare different existing GxE methodologies, providing a clear map of the GxE landscape.

Estimation Procedure

The core methodology of PIGEON is an extension of Linkage Disequilibrium (LD) score regression adapted for GxE analysis. The key feature of the PIGEON estimation procedure is its reliance solely on Genome-Wide Interaction Study (GWIS) and Genome-Wide Association Study (GWAS) summary statistics, avoiding the need for individual-level genotype data. The framework outlines two main objectives in polygenic GxE inference:

  1. Detecting GxE: Estimating the GxE variance component (\(\sigma_I^2\)), where a value greater than zero indicates the presence of GxE.
  2. Interpreting GxE: Estimating covariant GxE (\(\rho_{GI}\)) and Oracle Polygenic Score GxE (PGSxE), which quantifies the interaction between the environment and an individual’s true additive genetic component (PGS). The estimation of Oracle PGSxE is shown to be equivalent to estimating covariant GxE.

Analytical Advantages

The PIGEON method is robust to issues such as arbitrary sample overlap between GWAS and GWIS data and heteroskedasticity (non-constant residual variance across environments), which often complicate traditional GxE methods.

Results and Empirical Application

The paper demonstrates the effectiveness of PIGEON through extensive theoretical and empirical analyses:

  • Gene-by-Education Interaction: A quasi-experimental study of gene-by-education interaction was performed on health outcomes, showcasing PIGEON’s ability to analyze real-world policy-relevant exposures.
  • Gene-by-Sex Interaction: The method was successfully applied to quantify gene-by-sex interaction for a large set of 530 traits using data from the UK Biobank.
  • Gene-by-Treatment Interaction: PIGEON was used to identify genetic interactors that help explain the heterogeneity of treatment effects in a clinical trial focused on smoking cessation.

Conclusions and Recommendations

PIGEON provides an innovative and rigorous solution to long-standing challenges in polygenic GxE inference. By leveraging summary statistics and a unified variance component framework, the paper suggests a promising path that may fundamentally reshape analytical strategies in future GxE studies for complex human traits.