Preprint describes ultra-fast analytic method for multivariate GWAS using Genomic SEM

A preprint posted to bioRxiv describes a new analytic estimation framework for Genomic Structural Equation Modelling (Genomic SEM), an open-source R-based approach first introduced in 2018 that allows researchers to model the shared and distinct genetic architecture of multiple traits simultaneously. The Genomic SEM framework is widely used in psychiatric and complex-trait genetics to interrogate polygenic genetic relationships, identify loci with multi-trait effects, and guide multivariate genome-wide association study (GWAS) discovery.

The new implementation, described by the preprint authors, uses flexible analytic estimation — as opposed to numerical optimisation — to achieve what the authors characterise as a substantial reduction in computational time, without sacrificing statistical accuracy. This matters practically because multivariate GWAS analyses, which jointly model several phenotypes across millions of variants and large GWAS summary statistics files, are computationally intensive and have previously required substantial high-performance computing resources.

The preprint has not yet undergone peer review. Genetic Current notes that the Genomic SEM framework already has broad uptake in psychiatric genetics, neuroimaging genetics, and complex-trait research; an efficient implementation may lower the barrier to its use in educational and methodological training contexts as well as research applications.

This is a methods preprint and the results described have not been independently validated.