Evaluating and improving heritability models using summary statistics.

Area of Science:

• Population Genetics
• Statistical Genetics
• Genomics

Background:

• Debate exists on the optimal model for genomic heritability variation.
• Existing models include GCTA-LDMS-I, LD Score Regression's Baseline LD, and the LDAK model.
• Accurate heritability modeling is crucial for understanding complex trait genetics.

Purpose of the Study:

• To develop a statistical framework for assessing heritability models using genome-wide association study (GWAS) summary statistics.
• To compare the performance of existing heritability models and propose improvements.
• To estimate the genome-wide selection parameter (α) and investigate its impact on trait heritability.

Main Methods:

• Developed a statistical framework utilizing summary statistics from GWAS.
• Applied the framework to 31 large-scale human trait studies (average N=136,000).
• Incorporated features from the LDAK model to enhance the Baseline LD model.

Main Results:

• The Baseline LD model demonstrates greater realism compared to other current heritability models.
• Integrating LDAK model features significantly improves the Baseline LD model's accuracy.
• Strong evidence (P < 1x10⁻⁶) of negative genome-wide selection was detected for traits like height, blood pressure, and educational attainment.
• The impact of negative selection is more pronounced within functional genomic categories (e.g., coding SNPs, promoter regions).

Conclusions:

• The proposed framework offers a robust method for evaluating and refining heritability models.
• The enhanced Baseline LD model provides a more accurate representation of heritability architecture.
• Negative genome-wide selection significantly influences the heritability of complex human traits, particularly within functional regions.