Integrated systems approach identifies pathways from the genome to triglycerides through a metabolomic causal network

TL;DR

This study introduces a multi-stage causal network approach integrating genomics and metabolomics data to elucidate pathways influencing triglyceride levels, aiding mechanistic understanding of cardiovascular risk factors.

Contribution

The paper presents a novel method combining causal networks with GWAS to identify pathways from genome to triglycerides via metabolomics, enhancing mechanistic insights.

Findings

Identified a causal network over metabolomic levels using G-DAG.

Found significant effects of LRRC46 and LRRC69 mutations on metabolites affecting triglycerides.

Mapped pathways from FAM198B and C6orf25 to triglycerides through metabolites.

Abstract

Introduction: To leverage functionality and clinical relevance into understanding systems biology, one needs to understand the pathway of the genetic effects on risk factors/disease through intermediate molecular levels, such as metabolomics. Systems approaches integrate multi-omic information to find pathways to disease endpoints and make optimal inference decisions. Method: Here, we introduce a multi-stage approach to integrate causal networks in observational studies and GWAS to facilitate mechanistic understanding through identification of pathways from the genome to risk factors/disease via metabolomics. The pathways in causal networks reveal the underlying relationships behind observations, which do not play a significant role in more traditional correlative analyses, where one variable at a time is considered. Results: We identified a causal network over the metabolomic level using the genome directed acyclic graph (G-DAG), to systematically assess whether variations in the genome lead to variations in triglyceride levels as a risk factor of cardiovascular disease. We found LRRC46 and LRRC69 harboring loss-of-function mutations have significant effect on two metabolites with direct effects on triglyceride levels. We also found pathways of FAM198B and C6orf25 to triglycerides through indirect paths from metabolites. Conclusion: Integrating causal networks with GWAS facilitates mechanistic understanding in comparison to one-variable-at-a-time approaches due to accounting for relationships among components at intermediate molecular levels. This approach is complementary to experimental studies to identify efficacious targets in the age of big data sets.