Genomic Prediction and its Role in Livestock Improvement

Genetic improvement occurs when above-average selection candidates are chosen as parents of the next generation.  Genetic evaluation is the process of predicting the merit of selection candidates and in livestock improvement it has traditionally been done using recorded pedigree and performance information.  The most common approach involved fitting a mixed model that exploited assumed similarity among relatives to explain deviations in performance from cohort average in terms of breeding values and unexplained residual effects.  Similarity among animals was quantified in terms of covariances obtained from knowledge of the pedigree.  In contrast, genomic prediction derives covariances among animals by additionally accounting for genome-wide information on inherited variants. It promises more accurate predictions at young ages, which may allow greater intensities of selection and reduced generation intervals that could accelerate annual rates of genetic improvement.

Challenges in genomic prediction of livestock include the fact that there are usually more known variants than there are animals with performance information, animals of interest are closely related and therefore share large genome fragments, and that the routinely genotyped mutations do not include the causal variants.  Nevertheless, genomic predictions in large populations have already proven to be more accurate than pedigree-based parent-average predictions. Improving the accuracy of genomic prediction will require multi-disciplinary teams combining knowledge of biology, statistics, computing, and bioinformatics along with big data.