Founder-haplotype Mendelian randomisation method improves causal inference in Diversity Outbred mice
A preprint introduces Founder GLS-IVW, which leverages the eight-founder-strain structure of Diversity Outbred mice to improve Mendelian randomisation estimates and validate them against experimental knockouts.
A preprint posted to bioRxiv presents Founder GLS-IVW, a Mendelian randomisation (MR) method tailored to the Diversity Outbred (DO) mouse population. Standard summary-data MR methods summarise each genetic variant's association with an exposure as a single scalar, discarding information about which of the eight inbred founder strains contributes the effect. Founder GLS-IVW instead models the full pattern of founder-haplotype effects at each locus, using this richer signal for both causal estimation and instrument-validity diagnostics.
As a demonstration, the authors applied the method to the Qsox1 locus. Founder GLS-IVW recovered the direction of effect seen in published knockout mouse experiments — a direction that scalar MR approaches failed to capture correctly. Because DO mice allow causal hypotheses generated by MR to be tested directly through experimental intervention, the system offers a relatively rare opportunity to benchmark MR methodology against ground truth.
The work is primarily of interest to statistical geneticists and quantitative geneticists developing or applying MR methods, and to researchers using DO mice as a model system for complex trait genetics. It also has pedagogical value for courses covering instrumental variable methods and their assumptions. This is a preprint; it has not yet completed peer review.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-11-12Extending Mendelian randomisation to founder-haplotype data in Diversity Outbred mice