Preprint: DetectGxT software improves detection of gene-by-treatment interactions in molecular QTL studies

A preprint on bioRxiv introduces DetectGxT, an R software package designed to improve the statistical detection of gene-by-treatment (GxT) interactions in molecular quantitative trait locus (molQTL) studies. The tool targets a specific methodological problem: standard linear models used for interaction molQTL mapping are non-trivially misspecified when applied to molecular count data — such as RNA sequencing read counts — because they do not adequately account for the multiplicative (allelic additive on a log scale) relationship between genotype and phenotype.

By using nonlinear regression that models this relationship more accurately, the authors report increased statistical power to detect genuine GxT interactions without an accompanying increase in false-positive rates. The package is framed as applicable across multiple biological fields where understanding how genetic background modifies the response to an applied treatment — whether pharmacological, environmental, or developmental — is the primary question.

Interaction molQTL mapping is an increasingly important tool for dissecting context-dependent gene regulation and identifying mechanisms of drug response variation. For statistical geneticists, this is a methods contribution addressing a well-recognised but underserved modelling gap.

This is a preprint and has not yet been peer-reviewed. Findings should be treated as preliminary pending independent scrutiny and formal publication.