Effective number of genes and accuracy of genomic evaluations

Abstract

In this study we estimated the distribution of genetic variance for production and fertility traits across chromosomes, and used this information to calculate the effective number of genes associated to these traits. The estimated numbers of genes were then used in order to assess and to compare the expected accuracies of genomic evaluation using either GBLUP or BayesB. The regression of the proportion of genetic variance attributed to each autosome on its physical length fitted very well a linear relationship for all traits. The estimated effective number of genes ranged from ~400 (for fat percentage and non-return rate) to ~1000 (for milk yield, interval from calving to first insemination and days open). Our results provide evidence that a large number of genes is involved in the inheritance of milk production and fertility traits in dairy cattle. Expected accuracies of genomic predictions ranged from 0.76 to 0.87 for fertility traits, and from 0.83 to 0.92 for production traits. Expected accuracies of genomic evaluation were higher with BayesB for traits with lower number of QTL, and with GBLUP for larger number of QTL.