Integrative Pathway Analysis of SNP and Metabolite Data Using a Hierarchical Structural Component Model

Taeyeong Jung, Youngae Jung, Min Kyong Moon, Oran Kwon, Geum Sook Hwang, Taesung Park

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


Integrative multi-omics analysis has become a useful tool to understand molecular mechanisms and drug discovery for treatment. Especially, the couplings of genetics to metabolomics have been performed to identify the associations between SNP and metabolite. However, while the importance of integrative pathway analysis is increasing, there are few approaches to utilize pathway information to analyze phenotypes using SNP and metabolite. We propose an integrative pathway analysis of SNP and metabolite data using a hierarchical structural component model considering the structural relationships of SNPs, metabolites, pathways, and phenotypes. The proposed method utilizes genome-wide association studies on metabolites and constructs the genetic risk scores for metabolites referred to as genetic metabolomic scores. It is based on the hierarchical model using the genetic metabolomic scores and pathways. Furthermore, this method adopts a ridge penalty to consider the correlations between genetic metabolomic scores and between pathways. We apply our method to the SNP and metabolite data from the Korean population to identify pathways associated with type 2 diabetes (T2D). Through this application, we identified well-known pathways associated with T2D, demonstrating that this method adds biological insights into disease-related pathways using genetic predispositions of metabolites.

Original languageEnglish
Article number814412
JournalFrontiers in Genetics
StatePublished - 24 Mar 2022

Bibliographical note

Publisher Copyright:
Copyright © 2022 Jung, Jung, Moon, Kwon, Hwang and Park.


  • SNP
  • mGWAS
  • metabolite
  • multi-omics integration
  • pathway analysis


Dive into the research topics of 'Integrative Pathway Analysis of SNP and Metabolite Data Using a Hierarchical Structural Component Model'. Together they form a unique fingerprint.

Cite this