Fast probabilistic whitening method tackles linkage disequilibrium in ultra-high-dimensional genetic data
Gabriel Hoffman and colleagues at the Icahn School of Medicine describe a computationally efficient whitening transformation for removing linkage disequilibrium correlation structure from large-scale statistical genetics datasets.
Gabriel E. Hoffman, Christian P. Dillard, Kiran Girdhar, and Panos Roussos have published a methods paper in PLOS Genetics describing a fast probabilistic whitening transformation designed to handle the correlation structures that complicate analysis of ultra-high-dimensional genetic data.
A central challenge in statistical genetics is that variants located close together on the genome tend to be inherited together — a phenomenon known as linkage disequilibrium (LD). This correlation between nearby variants can distort downstream analyses, including applications that use GWAS summary statistics. Many existing methods address LD using reference panels, but scaling these approaches to very large datasets presents computational challenges.
The whitening transformation described in this paper offers a way to remove correlation structure from data before analysis, improving the validity of methods that assume independence between features or samples. The authors apply their approach to statistical genetics contexts where LD between regression summary statistics is a recognised complication. The work is a technical methods contribution most directly useful to statistical geneticists and computational genomics researchers developing or applying large-scale genetic association and functional genomics pipelines. It will also be of interest to educators covering the statistical foundations of genome-wide association studies.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary source PLOS Genetics · 2026-10-28Fast probabilistic whitening transformation for ultra-high dimensional genetic data