A general approach to testing for pleiotropy with rare and common variants

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) identify numerous genes linked to multiple phenotypes.
• Existing methods like multivariate regression and principal component analysis may not definitively confirm pleiotropy.
• The Pleiotropy Estimation and Test Bootstrap (PET-B) method is limited in its application to binary phenotypes, multiple traits, and rare variants.

Purpose of the Study:

• To develop and validate robust statistical approaches for formally testing pleiotropic relationships.
• To extend pleiotropy testing to accommodate categorical and quantitative phenotypes, multiple traits, and both common and rare genetic variants.
• To provide reliable methods for assessing genetic overlap in complex trait analysis.

Main Methods:

• Developed two novel approaches based on permuting phenotypes and comparing observed P-values to permuted P-values.
• Utilized the Hausdorff metric and a cutoff-based approach for P-value comparison relative to the origin (vector of zeros).
• Evaluated the methods through extensive simulation studies and applied them to the COPDGene dataset.

Main Results:

• The proposed methods demonstrated effectiveness in testing pleiotropy across diverse scenarios, including multiple phenotypes and variant types.
• The approaches successfully handled both categorical and quantitative traits, as well as common and rare variants.
• Application to the COPDGene study provided insights into the genetic architecture of chronic obstructive pulmonary disease.

Conclusions:

• The new permutation-based methods offer a flexible and powerful framework for testing pleiotropy in complex genetic studies.
• These approaches overcome limitations of existing methods, enabling broader application in genetic research.
• The findings contribute to a better understanding of genetic overlap and pleiotropy in human diseases.