Use Of Genomic SNP Information In Pig Breeding B. Nielsen * , T. Ostersen * , G. Su † , O.F. Christensen † , and M. Henryon * Introduction The DNA typing technology has now been developed to a stage where it is ready for im- plementation in calculation of breeding values as well as in the design of breeding schemes. It is expected that in the near future, pig breeding companies will increasingly use genomic information. Based on this, a combination of SNP information with classical BLUP evaluation seems ap- pealing. In future breeding programs the existing sources of information from pedigree infor- mation and performance data should be combined with the genomic information from SNP typing. Methods have been proposed to do so using all the data available (Aguilar et al., 2010; Christensen and Lund, 2010; Van Raden et al., 2009; VanRaden, 2008). The one-step approach suggested by Aguilar et al. (2010) and Christensen and Lund (2010) is expected to be superior to the simple index approach. However, before that approach is confirmed to be ready for implementation in routine genetic evaluation, a more simple animal model approach can be used as a transition. The objective of this paper is to combine genomic breeding values with traditional BLUP breeding values and to show how genomic information affects the predicted breeding values for both genotyped and non-genotyped animals. Material and methods The method presented here for combining genomic breeding values with traditional BLUP breeding values involved three steps. First, a prediction model associating SNP markers and traditional BLUP-breeding values was fitted on a training data set of old animals. Second, genomic breeding values were predicted for the young animals in the test data set using the fitted model from step one. In the third step the genomic breeding values of young animals in the test data set were included in a bivariate animal model in which trait 1 was phenotype recordings for all animals, and trait 2 was estimated genomic breeding values of the young animals. A genomic SNP model A prediction model associating SNP markers and traditional BLUP- breeding values was first fitted using a linear regression model with a random coefficient for each SNP marker. The overall distribution of all coefficients for the SNP markers was assumed to be Gaussian. Let x ij be the recorded gene content of animal i at SNP j such that x ij equals -1, 0, and 1 for the homozygote, the heterozygote, and the other homozygote, respectively. Then let y i be * Pig Research Centre, Axelborg, Axeltorv 3, DK-1609 Copenhagen V, Denmark † Aarhus University, Faculty of Agricultural Sciences, Department of Genetics and Biotechnology, Denmark