pIVW

pIVW estimates causal effects in Mendelian randomization by using genetic variants as instrumental variables from summary-level GWAS data while mitigating bias from weak instruments and balanced horizontal pleiotropy.

Key Features:

• Causal estimation: Provides an inverse-variance weighted estimator framework to estimate the causal effect of an exposure on an outcome using genetic variants as instrumental variables (IVs).
• Penalization for weak instruments: Incorporates a penalization approach that adjusts the IVW estimator to prevent the denominator from approaching zero and reduce bias from weak instruments.
• Adjustment for balanced horizontal pleiotropy: Modifies variance estimation to account for balanced horizontal pleiotropy where genetic variants affect multiple traits independently of the exposure-outcome pathway.
• Extension of dIVW: Builds upon the debiased IVW (dIVW) estimator and, under regularity conditions, achieves smaller bias and variance compared to dIVW.
• Validation via simulation: Performance and statistical properties have been evaluated and supported through extensive simulation studies.
• Use of summary-level GWAS data: Operates on summary-level genome-wide association study (GWAS) data for MR analyses.

Scientific Applications:

• Mendelian randomization analyses: Applied to infer causal relationships between exposures and outcomes in the presence of unmeasured confounding using genetic IVs.
• Obesity-related exposures and COVID-19 outcomes: Used to investigate causal links between obesity-related exposures and coronavirus disease 2019 (COVID-19), including associations such as hypertensive disease with increased risk of hospitalized COVID-19.
• Cardiometabolic and vascular traits: Identified associations of peripheral vascular disease and higher body mass index with elevated risks of COVID-19 infection, hospitalization, and critical illness.

Methodology:

Implements a penalized adjustment to the IVW estimator to stabilize the denominator, applies variance adjustment to account for balanced horizontal pleiotropy, extends the debiased IVW (dIVW) approach, and validates performance via simulation studies using summary-level GWAS data and genetic variants as IVs.