Practical implications for genetic modeling in the genomics era.

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Bioinformatics

Background:

• Genetic models are crucial for estimating breeding values and predicting animal performance.
• Historical data includes phenotypes and pedigrees, while recent data incorporates extensive genomic information.

Purpose of the Study:

• To provide accurate and timely predictions of future animal performance for breeders.
• To explore the complexities of genetic modeling, including environmental effects, interactions, and nonadditive inheritance.

Main Methods:

• Utilizing diverse data sources: phenotypes, pedigrees, and genomic data (SNPs, QTLs, haplotypes).
• Implementing various modeling approaches: linear, nonlinear, single-step, and multitrait models.
• Addressing computational challenges with large genomic datasets by selecting marker subsets.

Main Results:

• Genomic data has rapidly expanded, with millions of variants available.
• Current computations with ~60,000 markers take a few days.
• Genomic preselection can introduce bias in traditional models, though estimates are small in current US data.
• Single-step models integrate pedigree and genomic data, but require approximations for computational efficiency.

Conclusions:

• Models incorporating inbreeding, heterosis, dominance, and interactions can enhance mating predictions.
• Genomic reliabilities reach approximately 70% for highly heritable traits.
• Clear explanation of model workings is essential for breeders to confidently apply predictions.