This paper explores the potential benefits of family-based biobank sampling, a strategy that intentionally oversamples close genetic relatives rather than relying on random population-based recruitment. We argue that family-based biobanks can enhance causal inference in genetic studies, reduce confounding effects from shared environmental and demographic factors, and unlock new insights across multiple research domains. Given the limitations of traditional biobanks, this perspective highlights how integrating family-based designs could advance our understanding of genetic influences on health and disease, despite the added complexity and cost.
(Nature, 2024)
This article showed for the first time that estimates of SNP-heritability are upwardly biased for traits for which there is assortative mating (i.e., on traits that are correlated between mates, and when that correlation is due to matching based on the phenotype). For example, mates are similar on height, and this leads to a ~20% inflation of SNP-heritability estimates on height. This paper was the precursor to a subsequent paper by Border that showed potentially more serious biases on genetic correlations due to cross-trait assortative mating.
(Nature Communications, 2022)
This paper won the Fulker Award for best paper published in Behavioral Genetics in 2021. In it,Ìýwe introduce a model (SEMPGS) that demonstrates how to use polygenic scores derived from transmitted and non-transmitted parental haplotypes to giveÌýan unbiased estimate of the variation attributable to the environmental influence of parents on offspring, even when the polygenic score accounts for a small fraction of trait heritability.ÌýBy utilizing structural equation modeling techniques developed for extended twin family designs, our approach provides a general framework for modeling polygenic scores in family studies and allows for various model extensions that can be used to answer old questions about familial influences in new ways.
(Behavior Genetics, 2021)
This article showed that genomic similarity is trivially inflated by assortative mating, contrary to several recent publications showing increased similarity among mates or friends. Our paper argues that such inflated similarity is probably due to uncontrolled population stratification instead.
(Behavior Genetics, 2020)
This article used very large sample sizes to show somewhat definitively (in my opinion) that previous positive "candidate gene" findings on depression (of which there are many hundred) were almost certainly false positive findings. Our paper further calls into question tests on candidate gene hypotheses for most other traits. We argue that hypothesis-free genome-wide scans, which are about as cheap as candidate gene assays, produce much more reliable results and that the traditional candidate gene approach should be abandoned.
Ìý
(The American Journal of Psychiatry, 2019)
This article compared multiple approaches for estimating SNP-heritability, and introduced a variation on previous approaches (GREML-LDMS-I) that is substantially more accurate.
