Multivariate estimation of factor structures of complex traits using SNP-based genomic relationships.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Estimating heritability and genetic correlation from genome-wide single-nucleotide polymorphism (SNP) data is crucial in genetic studies.
• Existing methods can be computationally intensive for large datasets and multiple traits.
• Multivariate Genomic-Relatedness-Based Restricted Maximum Likelihood (MGREML) was previously developed for efficient estimation.

Purpose of the Study:

• To extend MGREML to fit and test user-specified factor models.
• To maintain computational efficiency while incorporating factor modeling capabilities.
• To enable advanced genetic architecture analysis using SNP data.

Main Methods:

• The extended MGREML method was applied to simulated datasets.
• Factor models were fitted and compared to nested models using real data (height and BMI).
• The method's performance was evaluated for statistical consistency and computational cost.

Main Results:

• Simulations demonstrated that the extended MGREML provides consistent estimates and valid inferences for factor models.
• The method achieves low computational cost, handling 50 traits and 20,000 individuals in under an hour on standard hardware.
• Real data analysis successfully illustrated the estimation and testing of a factor model.

Conclusions:

• The enhanced MGREML facilitates the estimation and inference of multivariate factor structures with high computational efficiency.
• This advancement enables researchers to perform structural equation modeling on genetic data.
• Researchers can now specify, estimate, and compare custom genetic factor models using SNP data via MGREML.