Summary: Norwegian breeding schemes have traditionally been based on large-scale collection of practical data. The advance of novel automated electronic recording systems further increases the data volumes coming from practical herds. However, these practical data come often with imperfections and many missing records. In the context of modern genomic selection breeding schemes, especially ungenotyped animals and thus missing genotypes cause biases and inaccuracies of breeding value estimates. A first major objective of the current project is to develop genetic evaluations for cattle and pigs that make seamless use of genotyped and ungenotyped animals, and other patterns of missing data. A second major objective is to develop algorithms that can handle massive amounts of practical records, including the ever-increasing numbers of animals genotyped with ever-increasing marker density. Also, a software package implementing the novel algorithms will be developed. A third major aim is to combine records on practical and elite breeding animals to pinpoint important genomic regions for the traits of interest, in order to maximize the accuracy of the genomic breeding value estimates. Statistical, animal breeding and machine learning approaches will be combined to tackle these objectives. The above challenges are faced by both the cattle and pig breeding industries and they thus join forces here to address them. By an increasing use of practical information, GENO and Norsvin aim to genetically improve the characteristics of the animals that are important under practical circumstances. Hence, making the animals better adapted to perform under practical conditions and thereby improve the sustainability of the cattle and pig production sectors.
Project leader: Janez Jenko
Institution: GENO SA