A simulation study on the detection of causal mutations from F2 experiments.

A simulation study has been performed to evaluate the power and the rate of false positives for the detection of causal mutations under two different models of analysis. We used an F2 design generated from an F0 population of five sires of line 1 and 40 dams of line 2 to produce an F1 population of 10 sires and 80 dams. Two different locations of the causal mutation and several frequencies of the mutations in the parental populations were considered. The first model included only the genetic configuration of the mutation, while the second model also included the probability of line origin given the neutral markers. Both models performed well when the mutation at the candidate gene was the causal mutation, although a greater power was obtained using the first model, because of its relative simplicity compared to the second one. However, when the candidate gene mutation was a neutral mutation, the second model presented a lower rate of false positives than the first. Moreover, in some cases the second model allowed distinction between the neutral and the causal mutation. The F2 design has a great power to detect quantitative trait loci provided by linkage disequilibrium, but also makes it difficult to discriminate between causal and neutral mutations. Therefore a high percentage of false positives can be expected. The limitations of F2 designs for discriminating between neutral and causal mutations are discussed.