Genomic prediction using mCADD scores as prior information in a mouse population

Area of Science:

• Genomics
• Quantitative Genetics
• Bioinformatics

Background:

• Standard genomic prediction (GP) models assume equal SNP contribution to genetic variation.
• Biological annotations can refine GP models by identifying more informative SNPs.
• Combined Annotation Dependent Depletion (CADD) scores predict the deleteriousness of genetic variants.

Purpose of the Study:

• To evaluate the benefit of using CADD scores for selecting or weighting SNPs in GP.
• To determine if CADD-based SNP selection/weighting improves prediction accuracy in a mouse dataset.

Main Methods:

• Analyzed 10 traits in 835 mice from the Diversity Outbred (DO) population.
• Used CADD scores (CADD-SNP and CADD-window) to select or weigh SNPs.
• Employed five GP models (GBLUP, BayesA, BayesB, BayesC, BayesR) and a weighted GBLUP model.

Main Results:

• Selecting the top 40% of SNPs based on CADD-window scores improved prediction accuracy for at least 6 out of 10 traits.
• CADD-SNP based selection did not improve accuracy.
• Weighting SNPs using CADD scores showed minimal impact on prediction accuracy.

Conclusions:

• SNP selection using CADD-window scores enhances genomic prediction accuracy, particularly for certain traits and GP models.
• The effectiveness of CADD-based SNP selection varies depending on the trait and the GP model used.
• Using CADD scores for weighting SNPs does not improve prediction accuracy.