Generating a robust statistical causal structure over 13 cardiovascular disease risk factors using genomics data.

Area of Science:

• Biomedical Sciences
• Genetics
• Epidemiology

Background:

• Understanding causal relationships is crucial in biomedical sciences.
• Directed Acyclic Graphs (DAGs) model system influences, but directionality can be ambiguous in observational data due to Markov equivalency.
• Exogenous data like genotypes can resolve causal directionality.

Purpose of the Study:

• To develop a robust statistical causal network among cardiovascular disease risk factor phenotypes.
• To utilize genome-wide SNP data and principal component analysis for inferring causal relationships.
• To identify key risk factors and potential intervention targets for cardiovascular diseases.

Main Methods:

• Applied principal component analysis (PCA) to 590,020 SNP genotypes from 1596 individuals.
• Used PCA-derived components to construct a genome-informed Directed Acyclic Graph (GDAG) representing causal relationships among 13 cardiovascular disease risk factors.
• Analyzed the GDAG to understand information flow and identify predictors and intervention targets.

Main Results:

• Successfully generated a robust statistical causal network (GDAG) among 13 cardiovascular disease risk factor phenotypes.
• Identified specific pathways and direct/alternative influences among risk factors.
• Highlighted Body Mass Index (BMI) as a significant factor influencing multiple other risk phenotypes.

Conclusions:

• The developed method effectively uses genomic information via PCA to establish causal networks among complex phenotypes.
• The identified causal network provides insights into the interplay of cardiovascular disease risk factors.
• BMI emerged as a critical node, suggesting it as a promising target for cardiovascular disease prevention interventions.