Research recently published in Genome Medicine shows that a widely-accepted model for predicting genetic risk of disease is not realistic when it is applied to current human disease data. Three other models provide a better fit to these data but are indistinguishable from each other.
Naomi Wray and Michael Goddard, from Queensland Institute of Medical Research and the University of Melbourne, examined mathematical models which integrate the disease risks from several genomic loci to determine the overall risk to an individual.
The unconstrained Risch model, which is commonly used in theoretical studies, was rejected as being unrealistic according to current human complex disease parameters. However, the CRisch, Odds, and Probit models were all compatible with these observed data.
The authors suggest that it will not be possible to distinguish between these last three models until more of the genetic variance causing human traits can be understood. As we learn more about the many kinds of mutations that are associated with genomic disease, so we can evaluate and model the ways that they interact to cause a phenotype.
Assistant Editor, Genome Medicine