A penalized linear mixed model for genomic prediction using pedigree structures.

Area of Science:

• Genetics
• Genomic Prediction
• Cardiovascular Research

Background:

• Genetic Analysis Workshop 18 (GAW18) focused on evaluating genomic prediction accuracy.
• Diverse pedigree structures were available for prediction analysis.
• Systolic blood pressure (SBP) prediction is crucial for cardiovascular health management.

Purpose of the Study:

• To investigate the utility of combining pedigree information with single-nucleotide polymorphism (SNP) data for predicting SBP.
• To assess the impact of different data types (SNP array vs. sequencing) on prediction accuracy.
• To determine if incorporating low-frequency variants improves SBP prediction.

Main Methods:

• Utilized GAW18 data, including pedigree structures and SNP genotypes from array and sequencing data.
• Developed prediction models for SBP incorporating pedigree information and SNP data.
• Compared prediction performance using different genetic data sources (common variants vs. common plus low-frequency variants).

Main Results:

• Pedigree information alone yielded unsatisfactory SBP prediction power.
• Integrating SNP genotypes substantially improved prediction accuracy.
• The improvement was most pronounced when a few SNPs had larger effect sizes.
• Including low-frequency variants from sequencing data did not enhance prediction accuracy compared to common variants alone.

Conclusions:

• Combining pedigree and SNP genotype data is superior to using pedigree alone for SBP prediction.
• Genomic prediction accuracy for SBP benefits from incorporating SNP data, especially variants with larger effects.
• Low-frequency variants do not appear to improve SBP prediction accuracy in this context.