A new statistical framework for genetic pleiotropic analysis of high dimensional phenotype data

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Traditional genetic pleiotropic analyses struggle with high-dimensional phenotype and genotype data.
• Existing methods are limited in examining relationships between common variants and numerous phenotypes.

Purpose of the Study:

• To develop a new framework for genetic analysis of multiple phenotypes using sparse structural equation models (SEMs).
• To extend SEMs to sparse functional SEMs, incorporating both common and rare variants.
• To address high-dimensional data challenges using functional data analysis and ADMM techniques.

Main Methods:

• Development of sparse functional structural equation models (SEMs).
• Integration of functional data analysis and ADMM techniques for dimensionality reduction and computational efficiency.
• Application to exome sequence data from the NHLBI's Exome Sequencing Project (ESP).

Main Results:

• Simulations show higher power for detecting causal genetic pleiotropic structures compared to existing methods.
• Gene-based pleiotropic analysis demonstrated increased power over single variant-based analysis.
• Identified a network of 137 genes connected to 11 phenotypes in the ESP data, with 114 genes showing pleiotropic effects.

Conclusions:

• Sparse functional SEMs effectively incorporate common and rare variants, with ADMM enabling efficient penalized SEM solutions.
• The model allows joint inference of genetic architecture and causal phenotype network structure.
• Proposed methods exhibit superior power in detecting true causal genetic pleiotropic structures.