Causal inference for heritable phenotypic risk factors using heterogeneous genetic instruments.

Area of Science:

• Genetics
• Epidemiology
• Biostatistics

Background:

• Genome-wide association studies (GWAS) demonstrate extreme polygenicity in complex traits.
• The
• all genes affect every complex trait
• phenomenon complicates Mendelian Randomization (MR) studies.
• Existing MR methods require reevaluation to accommodate pervasive horizontal pleiotropy and heterogeneous effect sizes.

Purpose of the Study:

• To propose a comprehensive framework, GRAPPLE, for analyzing causal effects with heterogeneous genetic instruments.
• To identify pleiotropic patterns from genetic data.
• To improve causal inference in the presence of complex genetic architectures.

Main Methods:

• Developed the GRAPPLE framework utilizing GWAS summary statistics.
• Incorporated methods to efficiently use both strong and weak genetic instruments.
• Enabled detection of multiple pleiotropic pathways, causal direction determination, and multivariable MR for confounding adjustment.

Main Results:

• GRAPPLE successfully analyzes causal effects of risk factors with heterogeneous genetic instruments.
• The framework can identify complex pleiotropic patterns.
• Applied to blood lipids, BMI, and systolic blood pressure, revealing new causal relationships and pathways for 25 disease outcomes.

Conclusions:

• GRAPPLE provides a robust framework for causal inference in polygenic traits.
• It enhances understanding of complex trait etiology by accounting for pleiotropy.
• Offers new insights into the causal relationships between key risk factors and various diseases.