Project Number: 8042-31000-002-012-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Aug 1, 2019
End Date: Jul 31, 2020
To 1) test algorithms that use detailed prior information about individual variants for improving genomic prediction, and 2) develop methods to fit interactions of variants with time (birth date) or with overall breeding value to test if each marker is becoming more important or less important as the population changes. Tests will be applied to simulated data and then to economically important traits of US dairy cattle. Strategies that emphasize causal variants and account for gene interactions should better predict future performance.
Selected causal alleles, sequence variants, and markers are now included in genotyping arrays. Genomic predictions could give the selected variants extra prior weight instead of equal weight to all. Accuracy of variant weighting strategies will be compared using time truncated phenotypes to predict more recent phenotypes. Effects of markers and causal alleles may change across generations due to epistasis, recombination, allele frequency, or genetic selection. Models could efficiently include interactions by fitting directional genotype by genotype effects (a subset of epistasis) using only n instead of n squared interactions to better predict future breeding values. Software (MMAP) already developed by the Cooperator will be revised to include the directional epistasis.