the question of whether information on genotype would be useful for prediction of risk. Moreover, risk alleles have been discovered mainly by using case-control studies, which, although efficient for the discovery of disease associated single nucleotide polymorphisms, are suboptimal for evaluating their performance as predictors of risk. This is optimally assessed by genotyping the genetic markers in prospective studies of initially healthy people in whom cases of incident disease accrue over time. However, the study designs, the number of disease associated single nucleotide polymorphisms evaluated, the metrics used to evaluate predictive performance, and the methods used to compare the performance of genotype with non-genetic risk factors have been variable. Despite this, genetic tests for the prediction of type 2 diabetes have already become available in the commercial sector (for example, www.23andme.com and www.decodeme.com/).
