Opportunities and computational challenges in large-scale whole-genome sequencing data analysis

Area of Science:

• Animal Genetics
• Bioinformatics
• Quantitative Genetics

Background:

• Genomic selection (GS) has been a vital tool in animal breeding for approximately 15 years, initially adopted by the dairy cattle industry.
• Advancements in genotyping technologies have led to increased adoption and larger genomic datasets, with full sequence data anticipated to replace SNP chips.

Purpose of the Study:

• To review methods and computational approaches for using sequence data in genomic prediction.
• To assess the impact of methods and model assumptions on prediction accuracy with sequence data.
• To discuss the modeling, development, applicability, and computational requirements for sequence data in GS.

Main Methods:

• Review of existing literature on genomic selection methods and computational approaches for sequence data.
• Analysis of the impact of different modeling strategies and assumptions on genomic prediction accuracy.
• Discussion of computational resources required for handling large-scale genomic data.

Main Results:

• Sequence data theoretically offers complete genetic variability but shows limited practical benefit over medium/high-density SNP panels for genomic prediction accuracy in cattle.
• Small effective population sizes (Ne) in cattle lead to long haplotype blocks, making it difficult to pinpoint causal variants within them.
• Medium-density SNP panels are effective in tagging these blocks, achieving high prediction accuracy with large datasets in cattle breeds.

Conclusions:

• Sequence data is best utilized for identifying causal variants to enhance prediction accuracy and stability, rather than direct use in genomic prediction.
• Optimal strategy involves using a subset of informative markers and appropriate models for genomic evaluation, especially with high-density data.
• Novel methods focusing on trait-specific markers could leverage sequence data by linking individuals through functional variants, requiring further research.