Sidebar
Still Missing
Eric Turkheimer
Research in Human Development, 2011, 8:3-4, 227-241
DOI: 10.1080/15427609.2011.625321
The missing heritability problem refers to the gap between heritability estimates for complex human traits based on quantitative genetics and the small magnitude and unreliability of contemporary molecular genetics, especially genome wide association studies. The author reviews the origins of the missing heritability problem and considers research that has attempted to resolve it by quantifying the joint explanatory power of multiple genetic loci, rather than considering their effects one at a time. Although this program has made an important contribution to understanding the role of genetics in the development of complex behaviors, it does not resolve the missing heritability problem.
This paper is a somewhat personal view of the “missing heritability” problem written by a prominent researcher from the population genetics tradition. At least as interesting are his views of the state of heritability research:
A final concern about the success of quantitative genetics is that it has been almost too successful. It was one thing when all the obvious candidates for genetic etiology [causation] were demonstrated, one after another, to be heritable: height of course, followed by weight or body mass index, intelligence and personality, schizophrenia and depression. But, with all the usual subjects exhausted, and many dissertations left to write, people conducted twin studies of less likely candidates, and to a troubling degree these all came out to be heritable as well. How much television children watch is heritable (Plomin, Corley, DeFries, & Fulker, 1990). Political attitudes are heritable (Alford, Funk, & Hibbing, 2005). Divorce is heritable (McGue & Lykken, 1992). There has been no end to it, especially as twin studies have expanded into the yet-to-be exhausted areas of economics (risk preference: Zhong et al., 2009) and political science (voting: Fowler, Baker, & Dawes, 2008)
The Missing Heritability Problem is the failure of many interpretations of GWAS results to generate anything like the level of genetic influence as predicted by heritability. It is rare for individual genetic markers in GWAS to predict more than 1% of variation in height, with the sum across markers not exceeding 10% or so, as compared to the 80% heritability of height. Although this mismatch clearly indicates a problem somewhere, Turkheimer favors the view that the problem lies with GWAS interpretation, and the genetic investigators' unfounded expectations that a few genes would explain a large part of the variance.
As of 2013 this issue is still far from settled, with others proposing that heritability estimates are too high because of non-additive gene interactions. However Turkheimer also argues that, according to a new methodology which is a hybrid of the GWAS and population genetics approaches, that the gap has been closed to a substantial degree, and that taken as a whole, these genomic results actually validate the heritability results, rather than refuting them.
The Visscher et al. (2010) program should drive a stake through the heart of a classical line of argument against classical behavioral genetics and its attendant statistical assumptions. Nevertheless, it is difficult to see how [this] will make much of an impact on the more contemporary problem, which is that quantitative genetics, despite demonstrating the universality of heritability, has failed to offer much in the way of etiological insight into complex behaviors, and moreover that the very ubiquity of heritability has made it problematic to differentiate between heritable phenotypes that have genetic mechanisms and those that do not.
He argues that the failure to discover any strong genetic markers for highly heritable traits (such as height) is because these traits are caused by the cumulative effects of thousands of genes, which is the inheritance model favored by heritability researchers. If this is true, then there is not much hope for finding humanly comprehensible models of how genes cause these traits, even though the heritability research proves the genetic basis.
See Behavioral Genetics.
