CCB Seminar: The effects of cross-trait assortative mating on genetic correlation estimators

Dr. Noah Zaitlen, Assistant Professor, UCLA
Abstract: Positive genetic correlation estimates are commonly interpreted as evidencethat two traits share overlapping biological features. With the increasing avail-ability of large consortium and biobank datasets, considerable research activityhas focused on cataloging the genetic correlation structure of complex humantraits in order to better understand shared etiology and transdiagnostic riskfactors. In the present manuscript, we demonstrate that primary-phenotypiccross-trait assortative mating (xAM) induces substantial genetic correlationsbetween genetically orthogonal traits, which are then further overestimated by widely-used marker-based estimators. Then, using empirical cross-matephenotypic correlation and heritability estimates across an array of previously investigated traits, we demonstrate that xAM alone can account for substan-tial fractions of nominal genetic correlation estimates. For example, in the absence of pleiotropy, xAM would induce genetic correlation estimates 33%(95% CI: 32.6%−34.4%) as large as those observed between anthropometric and psychosocial phentypes in the UK Biobank. Further, we demonstrate that previously reported genetic correlation estimates between many pairs of psychiatric disorders are fully consistent with moderate degrees of xAM and diagnostic errors. Finally, we provide evidence xAM at the genetic level by demonstrating that cross-trait even/odd chromosome polygenic score correlations are strongly associated with cross-mate cross-trait phenotype correlations(adjustedR2=37%). Together, our results demonstrate that genetic correlation is an unreliable indicator of shared etiology for traits subject to xAM and that previously reports are likely to represent overestimates of the true genetic similarity between many phenotype pairs.
Bio: I did my undergrad in Math and CS in Berkeley. In my last year I started working with Lior Pacther and moved into genetics. I spent a year at the Centre National de Genotypage in Paris under Mark Lathrop. I did my PhD with Eleazar Eskin, first at UCSD and then at UCLA. I did a postdoc with Eran Halperin in Tel Aviv University and then with Alkes Price at the Harvard School of Public Health. I was faculty at UCSF for five years and moved to UCLA in 2019 when I am now. My lab does statistical methods development across a broad array of topics including population genetics, quantitative genetics, and molecular technology development.