Biometrical modelling in genetics: are complex traits too complex?

¹ Divison of Epidemiology, Norwegian Institute of Public Health, Norway and Section for Epidemiology and Medical Statistics, University of Bergen, Bergen, Norway
² Section for Epidemiology and Medical Statistics, Department of Public Health and Primary Health Care, University of Bergen, Norway

^* To whom correspondence should be addressed.

	Abstract

The field of traditional biometrical genetics uses mixed-effects models to quantify the influence of genetic and environmental factors on a biological trait, based essentially on estimating within-family trait correlations. Such analyses provide a useful preview of what may be discovered with the emerging full-scale genotyping strategies. However, biometrical analyses require unrealistically large sample sizes to obtain a reasonable precision, particularly for dichotomous traits. In addition, it may be very difficult to separate genetic and environmental effects because environmental correlations are poorly understood. We illustrate these and other difficulties using population-based cousins and nuclear family data for birth weight, collected from the Medical Birth Registry of Norway.