Factors affecting accuracy of estimated effective number of chromosome segments for small breeds

Description

Small dairy cattle breeds often produce less than mainstream breeds, which endangers their existence. Genomic selection (GS) may signicantly increase the genetic improvement per year and can be used to improve long-term
perspectives of small breeds by making them more lucrative for farmers. GS requires a sufciently large breed-specic reference population for genomic prediction (GP), which is the main obstacle for numerically small breeds.
This may be overcome by adding individuals from other breeds to the reference population, however, the benet strongly relies on relatedness between the breeds, which can be inferred through the effective number of chromosome
segments (Me). In fact, Me is the key parameter in determining the accuracy of GP. Our objective was to investigate the number of individuals needed to accurately estimate Me within and between populations. In addition, we studied
the effect of marker density on the estimates of within and between population M
e. Two populations of 5,000 individuals separated for 100 generations were simulated to reect selection history and LD structure of current domestic cattle breeds, using QMSim software. We tested ve sample sizes of 10, 50, 100, 500, 1000 individuals on the accuracy of within population Me
and combinations of these sample sizes from each breed on the accuracy of
between population Me. Me is presented as the average of estimates from 50 replicates. The initial number of SNPs was 720k. We selected every 2x-th SNP, where x ranged from 1 to 6, to test the effect of marker density. The Me was
computed as the variance of genomic relationships within or between populations. Our results show that at least 50 individuals are needed to accurately estimate within population Me, regardless of SNP density. Lower sample size resulted in significant oscillations of the estimates. However, SNP density was important for between population Me. For example, 45k SNPs underestimated between population Me value by 20% compared to 720k. Importantly, for estimates of between population Me, both populations need to have at least 100 genotyped individuals.

Period	28 Aug 2019
Event title	70th Annual Meeting of the European Federation of Animal Science (2019)
Event type	Conference
Location	Ghent, Belgium
Degree of Recognition	International