MOD-scTWAS method improves cell-type-resolved genetic association with complex traits using gene co-expression
A preprint introduces a module-based transcriptome-wide association method that jointly models genetically regulated expression across co-expressed gene sets, improving power to detect cell-type-specific trait associations from single-cell data.
Transcriptome-wide association studies (TWAS) aim to identify genes whose genetically regulated expression (GReX) is associated with a complex trait, bridging GWAS signals and gene function. The emergence of population-scale single-cell transcriptomic data offers the prospect of running TWAS at cell-type resolution — an advance that could help pinpoint the specific cellular contexts in which associated genes act. Existing single-cell TWAS methods, however, have shown limited predictive performance for GReX, partly because they model genes independently and fail to exploit correlated expression structure.
A preprint posted to bioRxiv on 29 September 2026 describes MOD-scTWAS, a module-based method that addresses this limitation by jointly modelling GReX for genes within co-expression modules. Starting from a generative model for single-cell gene expression, the framework borrows information across genes in the same module to improve GReX prediction, then propagates these predictions into cell-type-resolved TWAS analyses.
The authors benchmark MOD-scTWAS against existing approaches across a range of complex traits and report improvements in predictive accuracy and in the identification of cell-type-specific associations. The method is designed to scale to biobank-sized datasets. This is a preprint and has not yet undergone peer review (bioRxiv doi: 10.64898/2026.09.24.754034).
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-09-29MOD-scTWAS: Leveraging gene co-expression for single-cell transcriptome-wide association studies