Tittle: Quantile-Based Mediation Analysis Methods for Genetic Association Studies

Abstract: Causal mediation analysis provides a formal statistical framework for assessing direct effects of genetic variants in genetic association studies. However, conventional mediation analysis methods based on mean regression models may fail to capture heterogeneous causal mechanisms across the conditional distributions of the target and intermediate phenotypes. To address this limitation, we develop a sequential G-estimation method to estimate the controlled direct effect of a genetic marker at arbitrary quantiles of the target phenotype under a hypothesized directed acyclic graph. Our approach utilizes estimating functions to derive a robust sandwich variance estimator. The proposed method provides consistent estimation of controlled direct effects with valid uncertainty quantification and achieves nominal coverage probabilities while accommodating confounding between the intermediate and target phenotypes through observed covariates.

---

Location: SN-2067

Date and Time: Friday, Oct. 2 at 09:00 AM - 12:00 PM (NDT)

---