Authors: Evan Charney
Abstract: The search for genetic risk factors underlying the presumed heritability of all human behavior has unfolded in two phases. The first phase, characterized by candidate-gene-association (CGA) studies, has fallen out of favor in the behavior-genetics community, so much so that it has been referred to as a “cautionary tale.” The second and current iteration is characterized by genome-wide association studies (GWASs), single-nucleotide polymorphism (SNP) heritability estimates, and polygenic risk scores. This research is guided by the resurrection of, or reemphasis on, Fisher’s “infinite infinitesimal allele” model of the heritability of complex phenotypes, first proposed over 100 years ago. Despite seemingly significant differences between the two iterations, they are united in viewing the discovery of risk alleles underlying heritability as a matter of finding differences in allele frequencies. Many of the infirmities that beset CGA studies persist in the era of GWASs, accompanied by a host of new difficulties due to the human genome’s underlying complexities and the limitations of Fisher’s model in the postgenomics era.
Key Findings
- The two phases of the search for the underlying heritability of human behavior-reliant first on candidate gene association (CGA) studies and then on genome wide association studies (GWASs) and polygenic scores-have both been marred by failures stemming from a variety of causes.
- CGA studies, predominant from 1990-2010 and consisting of a small number of polymorphisms of a handful of genes, suffered from widespread failures of replication.
- The second phase, analyzing millions of numbers of single nucleotide polymorphisms (SNPs) that each contribute miniscule variations, has required massive sample sizes to detect significant findings. However, these too have not been replicated with consistency.
- Nevertheless, statistically significant or “lead” SNPs have been used to create further metrics known as polygenic scores (or PGS), which are treated as predictors of individual risk. Such scores for educational attainment, intelligence, income have no predictive value whatsoever.
- Moreover, these types of analyses are largely restricted to “whites of European ancestry” (WoEA). Whenever they are applied to non-WoEA populations, their R-squared values plummet; defenses of these results lead to dangerous presumptions about genetic homogeneity within and heterogeneity across racial groups.
