How to model temporal changes in nontargeted metabolomics study? A Bayesian multilevel perspective

Area of Science:

• Biochemistry
• Systems Biology
• Statistical Modeling

Background:

• Time series data analysis is crucial for understanding physiological mechanisms and pathological alterations.
• High-dimensional, collinear, and noisy data present significant challenges for time series mining and analysis.
• Integrating the time variable into complex biological datasets requires robust statistical methodologies.

Purpose of the Study:

• To develop and test a Bayesian multilevel modeling approach for time series metabolomics data.
• To model hierarchical data structures and account for various sources of variability in in vivo studies.
• To assess the impact of pathological conditions, such as cancer, on metabolomics profiles over time.

Main Methods:

• Application of Bayesian multilevel modeling to time series metabolomics data.
• Utilizing a multilevel linear model with a double exponential prior for treatment effects.
• Incorporating robustness to outliers and modeling different levels of random effects.
• Analysis of inter-rat and inter-occasion variability and metabolite signal correlations.

Main Results:

• The proposed model demonstrated effectiveness in handling complex metabolomics data.
• Treatment effects for most metabolites were near zero, indicating minimal impact of cancer on the metabolomics profile.
• Inter-rat variability ranged from 3-30% (median ~10%), and inter-occasion variability ranged from 0-30% (median ~5%).
• Approximately 36% of metabolites exhibited outlying data points, and complex correlations between metabolite signals were identified.

Conclusions:

• Bayesian multilevel modeling is a suitable tool for analyzing time series metabolomics data, particularly in pharmacokinetic studies.
• The study suggests limited alterations in metabolomics profiles due to cancer under the investigated conditions.
• The methodology effectively captures variability and correlations within complex biological systems.