Related Experiment Video
Updated: May 1, 2026
Large-Scale Multi-Omics Genome-Wide Association Studies Mo-GWAS: Guidelines for Sample Preparation and Normalization
Published on: July 27, 2021
Assessing multivariate gene-metabolome associations with rare variants using Bayesian reduced rank regression
Pekka Marttinen1, Matti Pirinen2, Antti-Pekka Sarin1
1Department of Information and Computer Science, Helsinki Institute for Information Technology HIIT, Aalto University, Esbo, Finland, Center for Communicable Disease Dynamics, Harvard School of Public Health, Boston, MA, USA Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Unit of Public Health Genomics, National Institute for Health and Welfare, Helsinki, Computational Medicine, Institute of Health Sciences, University of Oulu and Oulu University Hospital, Oulu, NMR Metabolomics Laboratory, School of Pharmacy, University of Eastern Finland, Kuopio, Finland, Department of Epidemiology and Biostatistics, MRC Health Protection, Agency (HPA) Centre for Environment and Health, School of Public Health, Imperial College, London, UK, Institute of Health Sciences, Biocenter Oulu, University of Oulu, Oulu, Department of Clinical Physiology, Tampere University Hospital and University of Tampere, Department of Clinical Chemistry, Fimlab Laboratories, University of Tampere School of Medicine, Tampere, Finland, Computational Medicine, School of Social and Community Medicine and the Medical Research Council Integrative Epidemiology Unit, University of Bristol, Bristol, UK, Department of Clinical Physiology and Nuclear Medicine, Research Centre of Applied and Preventive Cardiovascular Medicine, University of Turku and Turku University Hospital, Turku, Department of Chronic Disease Prevention, National Institute for Health and Welfare, Helsinki, Unit of Primary Care, Oulu University Hospital, Department of Children and Young People and Families, National Institute for Health and Welfare, Oulu, Finland, Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK, Hjelt Institute and Department of Computer Science, Helsinki Institute for Information Technology HIIT, University of Helsinki, Helsinki, FinlandDepartment of Information and Computer Science, Helsinki Institute for Information Technology HIIT, Aalto University, Esbo, Finland, Center for Communicable Dise
This study introduces a novel Bayesian method to analyze the genetic basis of complex traits, improving the detection of associations with rare variants. The approach identified two novel genes, XRCC4 and MTHFD2L, linked to lipoprotein profiles, with findings replicated across multiple cohorts.
Area of Science:
- Genetics
- Statistical Genetics
- Bioinformatics
Background:
- Genome-wide association studies typically analyze single nucleotide polymorphisms (SNPs) against univariate phenotypes.
- Current methods struggle with high-dimensional phenotypes and the impact of rare variants.
- Univariate testing with stringent cutoffs limits the discovery of complex genetic dependencies.
Purpose of the Study:
- To develop a statistical approach for assessing the impact of multiple SNPs on high-dimensional phenotypes.
- To enhance the detection of genetic associations, particularly those involving rare variants.
- To identify novel gene-phenotype associations using a robust statistical framework.
Main Methods:
- Bayesian reduced rank regression was employed to analyze associations between multiple SNPs and high-dimensional phenotypes.
- The method was validated using genome-wide SNP data and lipoprotein profiles from the Northern Finland Birth Cohort.
- Performance was compared against alternative statistical approaches.
Main Results:
- The proposed Bayesian method effectively combined information across multiple SNPs and phenotypes, proving suitable for rare variant analysis.
- Two novel genes, XRCC4 and MTHFD2L, were identified as associated with lipoprotein profiles.
- These findings were successfully replicated in independent cohorts (Cardiovascular Risk in Young Finns and FINRISK studies).
Conclusions:
- The Bayesian reduced rank regression approach offers a powerful tool for exploring complex genetic architectures.
- This method advances the ability to detect associations involving rare variants and high-dimensional traits.
- The identified genes provide new insights into the genetic regulation of lipoprotein metabolism.
