Estimation of the genetic Gaussian network using GWAS summary data

TL;DR

This paper introduces EGG, a novel method for accurately estimating genetic networks from GWAS summary data by removing biases, thereby improving interpretation of biological relationships and causal contributions among phenotypes.

Contribution

The paper presents EGG, a new approach that corrects bias in genetic network estimation from GWAS data, enhancing interpretability and causal inference.

Findings

EGG outperforms traditional estimators in simulations.

EGG effectively removes bias due to estimation errors and pleiotropy.

Real data analysis confirms EGG's superior performance.

Abstract

Genetic Gaussian network of multiple phenotypes constructed through the genetic correlation matrix is informative for understanding their biological dependencies. However, its interpretation may be challenging because the estimated genetic correlations are biased due to estimation errors and horizontal pleiotropy inherent in GWAS summary statistics. Here we introduce a novel approach called Estimation of Genetic Graph (EGG), which eliminates the estimation error bias and horizontal pleiotropy bias with the same techniques used in multivariable Mendelian randomization. The genetic network estimated by EGG can be interpreted as representing shared common biological contributions between phenotypes, conditional on others, and even as indicating the causal contributions. We use both simulations and real data to demonstrate the superior efficacy of our novel method in comparison with the traditional network estimators. R package EGG is available on https://github.com/harryyiheyang/EGG.