Integrative analysis of metabolomics and transcriptomics data: a unified model framework to identify underlying...

Area of Science:

• Systems Biology
• Statistical Genomics
• Metabolomics

Background:

• High-dimensional biological data, such as gene expression and mass spectrometry, require advanced models to understand biological systems.
• Existing methods often rely on prior knowledge, limiting their application to well-studied organisms.
• A need exists for versatile methods applicable across diverse organisms.

Purpose of the Study:

• To develop a statistical method for integrating gene expression and Liquid Chromatography-Mass Spectroscopy (LC-MS) data.
• To identify genes that control metabolite production using integrated high-dimensional datasets.
• To provide a flexible and computationally efficient framework for biological data analysis.

Main Methods:

• Utilized a statistical approach combining gene expression (DNA microarray) and LC-MS data from the same biological samples.
• Employed dimension reduction and variable selection techniques to handle high-dimensional data.
• Identified basis functions in mass spectrometry data and related their weights to gene expression to find gene-metabolite associations.

Main Results:

• Successfully identified genes regulating specific metabolite production.
• Demonstrated the method's effectiveness in linking gene activity to metabolite profiles.
• The framework accurately identifies regulatory genes provided signal variation exceeds noise variation.

Conclusions:

• The proposed statistical method offers a powerful tool for uncovering gene-metabolite regulatory relationships.
• This approach is particularly valuable for organisms lacking extensive prior biological information.
• The flexible and efficient framework facilitates deeper insights into biological systems through integrated data analysis.