Quantitative reconcilliation of GWAS and candidate gene findings: measurement error, nonlinearity, and artifactual results

Candidate gene and genome-wide investigations of complex traits and related endophenotypes are both fundamentally concerned with elucidating the biological pathways underlying individual differences in behavior. However, these... [ view full abstract ]

Candidate gene and genome-wide investigations of complex traits and related endophenotypes are both fundamentally concerned with elucidating the biological pathways underlying individual differences in behavior. However, these two methods have produced disparate literatures and have often reached contradictory conclusions about the genetic architectures of target traits. Previous attempts at reconciling these discrepancies have been largely theoretical, and both methods continue to gain popularity in the broader social sciences. In contrast, the present research uses rigorous quantitative methods to evaluate plausible explanations for the current disagreement in the literature: poor measurement, coarse phenotyping, and simplistic models in genome-wide studies versus publication bias, methodological errors, and non-replication in candidate gene work. Particularly, we investigate the extent to which the above arguments might explain previously published results, relying on both analytic results and reanalysis of genome-wide association study results. We aim to aid investigators in parsing the fragmented empirical literature and in selecting future lines of inquiry likely to advance our understanding of genetic architecture.