Integrating Multiple Correlated Phenotypes for Genetic Association Analysis by Maximizing Heritability

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Genetic studies often analyze correlated disease variables jointly to enhance the detection of causal genetic variants.
• Existing methods include Bonferroni correction, multivariate regression, and dimension reduction techniques like principal component analysis.

Purpose of the Study:

• To propose a novel method for constructing a maximally heritable (MaxH) phenotype by combining individual phenotypes.
• To increase the power to detect causal genetic variants by maximizing heritability at individual SNPs.

Main Methods:

• Constructing a MaxH phenotype as a linear combination of individual phenotypes, leveraging estimated heritability and co-heritability.
• Applying the method to genome-wide scans, as heritability and co-heritability estimation is performed only once.
• Comparing the MaxH phenotype's heritability and power against commonly used methods.

Main Results:

• Simulations indicate that the MaxH phenotype achieves theoretical heritability and power for large samples and two phenotypes.
• The MaxH phenotype demonstrates increased heritability and power compared to the individual phenotypes being combined.
• The approach shows practical relevance through an application to a genome-wide association study (GWAS) for chronic obstructive pulmonary disease.

Conclusions:

• The MaxH phenotype offers a powerful approach for genetic studies analyzing multiple correlated disease variables.
• The method is applicable to genome-wide scans and provides a practical tool for identifying genetic associations.
• Suggestions are provided for selecting optimal phenotypes for combination to maximize study power.