Is the use of formulas a reliable way to predict the accuracy of genomic selection?

Abstract Text: We studied 4 formulas for prediction of accuracy of genomic selection. Our objectives were to study the variation of accuracy depending on the parameters, and to compare observed accuracy in 13 references (145 values) to the accuracy given by formulas. The marginal distribution of accuracy was studied according to each parameter (range of values defined by the references). Then we compared accuracies predicted using formulas and observed accuracies (got by cross-validation for real data or correlation for simulations). We proved that the size of the reference population and the number of effective segments M_e have a major weight and that there were large differences between formulas. The formula which gave the best prediction of observed accuracy depended on M_e estimation given the effective size of population and so no generalization can be done, and no use of formulas to decide genomic plans.

Keywords: Genomic selection, Accuracy, Effective number of segments