Whole-genome regression models have become ubiquitous for analysis and prediction of complex traits. In human genetics, these methods are commonly used for inferences about genetic parameters. This is so despite the fact that some of the assumptions commonly adopted for data analysis are at odds with important quantitative genetic principles. In this article we develop theory that leads to a precise definition of parameters arising in regression models using genomic data. Our approach is framed within the classical quantitative genetics paradigm. We discuss how these parameters relate to statistical parameters, indicate potential inferential problems and provide a limited set of simulations where some statistical properties of likelihood-based estimates are assessed.