Xuanjun Fang

Hainan Provincial Key Laboratory of Crop Molecular Breeding, Hainan Institute of Tropical Agricultural Resources (HITAR), Sanya, 572025, Hainan, China
Author    Correspondence author
Tree Genetics and Molecular Breeding, 2026, Vol. 16, No. 1   
Received: 17 Mar., 2026    Accepted: 12 Apr., 2026    Published: 15 May, 2026

This is an open access article published under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Complex traits are typically characterized by substantial genetic correlation and pleiotropy. However, conventional single-trait GWAS frameworks are limited in their ability to distinguish true shared causal effects from spurious associations arising from linkage disequilibrium, sample structure, or mediated relationships. Here, we advance multi-trait analysis from a collection of methods to a unified statistical genetic framework centered on clearly defined estimands, establishing a hierarchical inference system spanning covariance structure, locus configuration, and association patterns. Within this framework, Genomic SEM characterizes cross-trait genetic covariance and latent shared factors at the structural level; PLEIO performs joint fine-mapping within local LD structure to resolve causal configurations at the locus level; and Primo decomposes multi-trait association patterns using Bayesian mixture modeling to quantify shared and trait-specific effects at the pattern level. These approaches correspond to distinct inferential layers and collectively form a progressive evidence chain from structural reconstruction to effect decomposition. Through simulation and empirical analyses, we systematically evaluate the bias–variance trade-offs of multivariate methods under varying genetic correlation, LD complexity, and sample overlap scenarios, delineating the conditions under which multi-trait models improve power versus inflate false positives. We further emphasize a multi-evidence framework integrating local genetic correlation, joint fine-mapping, colocalization, effect direction consistency, and cross-ancestry validation to distinguish true pleiotropy from spurious signals. Building on this theoretical foundation, we propose a structured “screen–validate–apply” workflow: screening trait sets via genome-wide and local genetic correlation, resolving shared architectures using PLEIO and Primo, validating consistency through colocalization and conditional analyses, and finally expanding shared signals via Genomic SEM for downstream network and functional interpretation. This framework is broadly applicable to both crop genetics and human disease studies, providing a systematic pathway from statistical association to mechanistic insight and translational application.

Multi-trait genetic analysis; Pleiotropy; Genetic correlation (rg); Genomic SEM; Joint fine-mapping; Bayesian mixture models; Colocalization; Causal inference