Genes or environment? The difficulties of disentangling these effects in human genetic data analysis

Biostatistical methodology to accommodate genotype by environment (G-E) interactions in planned breeding trials for animals and for plants (such as crop variety trials) is well-advanced. Incorporation of environmental effects into genetic models for human data is generally more problematic. For continuous traits, such as height, IQ, blood pressure and cholesterol level measurements (to name a few), there is a vast literature on approaches that can be traced back over 80 years. However, many of the more modern approaches to realistic biostatistical modelling have not penetrated recent analyses of such data. For complex traits, like autoimmune diseases, cancer, cardiovascular disorders, obesity, psychiatric disorders, for example, recent emphasis has been on mapping supposed susceptibility (genetic) loci. Massive enterprises have been undertaken, but the outcomes have been mixed, and usually only applicable for subforms of the disease under study. Nevertheless, there is currently great enthusiasm for undertaking linkage studies for these diseases. Here a brief overview of some of the difficulties in realistically disentangling environmental and genetic effects in models for human genetic data is given.