Methodological analysis reveals biases in heritability estimates from ultra-rare variants in UK Biobank data.

This methodological analysis evaluated biases inherent in estimating heritability from ultra-rare variants using data from 305,813 unrelated European-ancestry individuals in the UK Biobank. The exposure involved an analysis of 5,330,210 exome-sequenced singletons, which are variants observed only once. The primary outcome assessed was the impact of population stratification on singleton-based heritability estimates, while secondary outcomes included heritability estimates for 22 quantitative phenotypes after accounting for identified biases.

Results indicated that population stratification induces both upward and downward biases in heritability estimates. Furthermore, calibration of asymptotic standard errors from likelihood-based procedures was found to be generally mis-calibrated when traits were not normally distributed. Specific heritability estimates were calculated for several phenotypes, including the number of children (effect size 3.4%), peak expiratory flow (1.9%), red blood cell count (2.5%), white blood cell count (1.9%), and heel bone mineral density (2.4%). The analysis also noted that these estimates capture non-additive genetic effects.

Safety and tolerability were not reported as this was a methodological analysis rather than a clinical trial. Key limitations include the fact that reliable heritability estimates for ordinal and binary traits will likely require far larger sample sizes and improved methods, as confounding in these traits remains difficult to detect and correct. The study provides recommendations for robust inference of heritability from ultra-rare variants, emphasizing that association only exists and no causation is implied. The certainty of these findings is based on methodological analysis using theory.