Genes, environments, and traits: Can you predict when we are going to understand them?

As genomics has entered the ‘big data’ era over the past 2 decades, phenotypic prediction has become central to fields ranging from livestock production to human health. In the latest case, efforts have focused primarily on the development of polygenic risk scores (PRS), derived from genome-wide association (GWA) studies. As prediction methods gained popularity, earlier gene-mapping approaches have been largely perceived as a dead end. Gene-mapping experiments did not rapidly reach the high level of expectation they raised, and the incentives for further improving their methodologies soon decreased. It could be argued that the study of genotype-phenotype relationships has taken sides in the broader scientific debate between prediction and explanation (also referred as causation or understanding). But fortunately, a major obstacle to the search of complex, polygenic (or even more broadly, polyfactorial) bases of traits has recently been overcome. Technically, the Natural and Orthogonal InterActions (NOIA) model enables unbiased estimates of effects of complex genetic and environmental architectures of traits, even in datasets with severe departures from equilibrium genotypic frequencies. This achievement significantly increases our potential to reveal the factors underlying phenotypes and allows for a sounder interpretation and reliable comparisons of different mapping experiments on the same trait. Improving our understanding of the relationship among genes, environments, and traits may provide us with countless advantages. Thus, the potential of advanced genetic modeling for studying complex traits needs to be assessed in practice, particularly when evidence suggests that PRS may be approaching a structural limit.