Efficient estimation of SNP heritability using Gaussian predictive process in large scale cohort studies.

Area of Science:

• Genetics
• Statistical Genetics
• Computational Biology

Background:

• Estimating heritability from genome-wide SNP data is crucial for understanding genetic contributions to traits.
• Linear mixed models (LMMs) are commonly used but computationally challenging for large cohorts.
• High-throughput genetic data necessitates efficient analytical methods.

Purpose of the Study:

• To develop a computationally efficient method for heritability estimation in large-scale cohort studies.
• To approximate linear mixed models (LMMs) using genetic coalescence and Gaussian predictive processes.
• To provide a fast alternative for heritability estimation in genomics research.

Main Methods:

• Proposed PredLMM, a novel method approximating linear mixed models (LMMs).
• Leveraged concepts of genetic coalescence and Gaussian predictive processes.
• Analyzed computational complexity compared to existing LMM-based methods.

Main Results:

• PredLMM demonstrates substantially improved computational complexity over existing LMM methods.
• Theoretical analysis shows PredLMM's approximation aligns with established Gaussian process methods.
• Successfully applied PredLMM to estimate heritability for multiple quantitative traits in the UK Biobank cohort.

Conclusions:

• PredLMM offers a computationally feasible and accurate approach for heritability estimation in large genetic studies.
• The method provides a valuable tool for analyzing high-throughput genetic data.
• PredLMM facilitates the study of genetic architecture in large populations.