Analytical and computational solution for the estimation of SNP-heritability in biobank-scale and distributed...

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Heritability estimation quantifies genetic determination of complex traits.
• Biobank-scale datasets necessitate more powerful and efficient estimation methods.

Purpose of the Study:

• To develop and validate a novel method for estimating SNP-heritability in biobank-scale data.
• To address the computational challenges of analyzing large genetic datasets.

Main Methods:

• The study introduces RHE-reg, a method based on Haseman-Elston regression integrated with a fast randomization algorithm.
• An analytical solution was developed to balance computational cost and estimation precision.
• The method was tested on simulated data and applied to 81 quantitative traits from the UK Biobank.

Main Results:

• RHE-reg efficiently handles biobank-scale data, demonstrated by its application to nearly 300,000 UK Biobank individuals.
• Estimation on UK Biobank data took approximately 4.5 hours using 10 CPUs.
• The method showed accurate heritability estimation in both simulated and real-world (UK Biobank) datasets.
• RHE-reg was successfully extended to distributed datasets.

Conclusions:

• RHE-reg provides an accurate and computationally efficient solution for SNP-heritability estimation in biobank-scale studies.
• The developed software is released, facilitating broader application in genetic research.
• The method is suitable for analyzing large, distributed genetic datasets while maintaining privacy.