A collaboration between Leonardo Bottolo (Medical Genetics), Hélène Ruffieux, Elena Vigorito, Chris Wallace, Sylvia Richardson (all MRC Biostatistics Unit) with Benjamin Fairfax and Isar Nassiri (both MRC Weatherall Institute for Molecular Medicine, Oxford), has led to a scalable statistical tool to refine the detection and interpretation of genetic regulation.

Published in The American Journal of Human Genetics, the new fully automatic approach enhances genome-wide molecular quantitative trait locus (QTL) mapping, and in particular the discovery of genetic controls of gene expression, by leveraging epigenetic annotations at the variant level. The method, called EPISPOT, effectively couples simultaneous QTL analysis of thousands of genetic variants and thousands of molecular traits with a hypothesis-free selection of biologically interpretable annotations which directly contribute to the QTL effects.

EPISPOT supports flexible sparsity constraints for QTL and epigenetic mark discovery and scales well as the sample size, the number of molecular traits and the dimension of the epigenetic annotation increase thanks to its computationally efficient variational inference implementation. The R package is freely available at https://github.com/hruffieux/epispot. EPISPOT builds on ATLASQTL, a joint eQTL mapping approach published by the same team in 2020 in The Annals of Applied Statistics.