Given the importance of attractiveness across interpersonal contexts, studies that investigate the underlying genetics of facial attractiveness, such as the one reported by Hu and colleagues [9] in this issue, are invaluable but should be interpreted carefully, commensurate with the complexity of attractiveness as a phenotype. Although Hu and colleagues report considerably lower heritability estimates for facial attractiveness than a previous estimate [10], perhaps due to modest interrater reliability (S13 Fig in [9]), evidence of heritability suggests that searches for underlying loci associated with attractiveness may bear fruit. Datasets with genome-wide genetic data and rated facial attractiveness are rare and time-consuming to gather, and Hu and colleagues smartly leverage a large, pre-existing dataset. After testing 6 overlapping sets of facial attractiveness ratings, they find 1 SNP associated with rated facial attractiveness at a study-wide threshold, 1 SNP significant at genome-wide threshold, and 10 suggestively significant SNPs.

Through a series of enrichment tests, Hu and colleagues identify several correlations between attractiveness ratings and genes influencing other traits—namely, body mass index in females and lipid traits in males (Fig 4 in [9]). Indeed, this study manifests as an illustration of the ability of a large GWAS on a complex phenotype to identify genes related to its simpler component traits and correlates. The candidate genes identified for both significant results and nearly all suggestive results have entries in GWAS Catalog for traits related to attractiveness, including skin pigmentation and melanoma, body mass index (BMI), and the BMI-related phenotypes of height and waist–hip ratio (Table 1). Homogeneous skin coloration [11] and red and yellow tints [12] increase ratings of attractiveness cross-culturally, potentially due to the connection between these traits and perceptions of health and youth. The relationship between weight and attractiveness is demographically variable; for example, American men of European descent rate lower weights as more attractive, except in extremely low BMI ranges [13], whereas African American men are more likely to prefer heavier figures [14]. Hu and colleagues also identify candidate genes related to attractiveness that have been previously associated with facial morphology, possibly implicating facial traits (such as those contributing to youthful facial appearance) in perceptions of attractiveness [15]. One explanation for Hu and colleagues finding a genetic association between male-rated female attractiveness and BMI is a mediating relationship whereby the candidate gene (CDC42EP3) affects height, which directly influences BMI. Similarly, the genetic association between female-rated male attractiveness and lipid levels observed by Hu and colleagues could be explained by the previously identified impact of candidate genes CERS2 and ANXA9 on both high- and low-density lipoprotein cholesterol levels (Table 1). It is also possible that Hu and colleagues find different loci for male- and female-rated attractiveness because men and women seem to vary in the specific traits they perceive as attractive [16].

The results of this study point to underlying genetic architecture mediating attractiveness. In the future, careful multivariate studies testing the relative contribution of each associated locus to the component traits of attractiveness, and to attractiveness corrected for those traits, will help researchers unravel and interpret the genetic architecture of this important and complex phenotype. Of course, replication in the few other datasets possessing both genotype and attractiveness data will aid in validation and resolution of these results, and sequencing studies will help clarify the possibly functional variants at each locus and further explore their effect on attractiveness or its related components. Hu and colleagues also briefly mention signatures of selection on alleles associated with male facial attractiveness. This result is especially intriguing and brings up several avenues for future research. Do other secondary sex traits, such as vocal characteristics, show similar signatures of selection in males, indicating sexual selection among our male ancestors [17]? Importantly, does the selection pressure driving the strong relationship between allele frequency and male attractiveness reflect pressure upon the attractiveness per se, or upon related phenotypes, such as lipid metabolism? How do potential signatures of selection fit in with previous evolutionary hypotheses? If there are causative pathways between the associated loci and attractiveness, have cross-cultural variations in preference [18] led to population-specific allele variation at these candidate attractiveness loci?

When contemplating how to depict Helen of Troy, the 5th century BC painter Zeuxis recognized the challenge of identifying the features that define beauty [19]. This challenge remains, and understanding the biological factors that influence attractiveness is equally compelling and complex. Hu and colleagues bring forth a valuable initial foray into the genetic architecture of attractiveness and emphasize the intricate relationships between attractiveness and other visible traits.