Multivariable G-E interplay in the prediction of educational achievement

Abstract

Polygenic scores are increasingly powerful predictors of educational achievement. It is unclear, however, how sets of polygenic scores, which partly capture environmental effects, perform jointly with sets of environmental measures, which are themselves heritable, in prediction models of educational achievement.Here, for the first time, we systematically investigate gene-environment correlation (rGE) and interaction (GxE) in the joint analysis of multiple genome-wide polygenic scores (GPS) and multiple environmental measures as they predict tested educational achievement (EA). We predict EA in a representative sample of 7,026 16-year-olds, with 20 GPS for psychiatric, cognitive and anthropometric traits, and 13 environments (including life events, home environment, and SES) measured earlier in life. Environmental and GPS predictors were modelled, separately and jointly, in penalized regression models with out-of-sample comparisons of prediction accuracy, considering the implications that their interplay had on model performance.Jointly modelling multiple GPS and environmental factors significantly improved prediction of EA, with cognitive-related GPS adding unique independent information beyond SES, home environment and life events. We found evidence for rGE underlying variation in EA (rGE = .36; 95% CIs = .29, .43). We estimated that 38% (95% CIs = 29%, 49%) of the GPS effects on EA were mediated by environmental effects, and in turn that 18% (95% CIs =12%, 25%) of environmental effects were accounted for by the GPS model. Lastly, we did not find evidence that GxE effects collectively contributed to multivariable prediction.Our multivariable polygenic and environmental prediction model suggests widespread rGE and unsystematic GxE contributions to EA in adolescence.

Genetic factors explain a significant part of associations between adolescent well-being and the social environment

“…For example, if twin data are available, direction of causation models (if there are different modes of inheritance for the traits under study) [44] and genetic cross-lagged models [45] provide genetically sensitive methods for studying causality. In the absence of family data, one can still try to separate genetic from environmental effects if DNA data are available, for example by incorporating the effect of polygenic scores (scores that reflect individuals' genetic predisposition for a trait based on results from genome-wide association studies) in mediation models [46] or Mendelian randomization models [47]. From a research perspective, it is important that investigations into adolescent mental health correlates take into consideration that these associations might reflect a shared genetic liability.…”

From Genome-Wide to Environment-Wide: Capturing the Environome

“…Although an extensive, conclusive body of empirical evidence has shown that correlations between genetic factors and environments are commonplace, G×Es remain to be reliably demonstrated (Duncan & Keller, 2011; Manuck & McCaffery, 2014; Tucker-Drob & Bates, 2016). For example, a recent large-scale study that tested GrEs and G×Es across multiple genotypes, phenotypes, and environments in the prediction of educational achievement observed systematic GrEs but concluded that the contributions of G×Es were random, weak, and negligible (Allegrini et al, 2020). Compared with detecting direct effects of genotypes on phenotypes (i.e., GrE), identifying G×Es is statistically more demanding because to be adequately powered, interaction models require much larger sample sizes than tests of direct effects (Duncan & Keller, 2011; Moffitt et al, 2006).…”

Putting psychology to the test: Rethinking model evaluation through benchmarking and prediction

“…One notable example is educational achievement. While early GWAS-based models explained only around 2% of the variance (Rietveld et al, 2013), current models explain as much as 20% (Allegrini et al, 2019;Lee et al, 2018;Selzam et al, 2018;von Stumm et al, 2020). Similar improvements in levels of predictive performance can be observed for other complex phenotypes, such as personality (Power & Pluess, 2015;Vinkhuyzen et al, 2012) and psychiatric traits (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Visscher et al, 2012;Wray et al, 2014).…”