SMILE: systems metabolomics using interpretable learning and evolution.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Metabolomics is crucial for understanding disease by studying small molecular metabolites.
• Machine learning (ML) is increasingly used in metabolomics for its predictive power.
• The 'black-box' nature of ML hinders interpretation, which is vital in biomedical research.

Purpose of the Study:

• To develop a novel computational framework for supervised metabolomics data analysis.
• To address the challenge of interpretability in ML models within metabolomics.
• To identify influential metabolites and their interactions associated with disease.

Main Methods:

• Proposed Systems Metabolomics using Interpretable Learning and Evolution (SMILE) framework.
• Utilized an evolutionary algorithm for learning interpretable predictive models.
• Developed a web application with a graphical user interface for analysis and visualization.

Main Results:

• SMILE successfully identified influential metabolites associated with Alzheimer's disease (AD).
• The framework generated interpretable predictive models for AD.
• Demonstrated the utility of the SMILE web application using AD metabolomics data.

Conclusions:

• SMILE enhances the understanding of the metabolic background of AD through interpretable models.
• The framework contributes to addressing the need for interpretability and explainability in bioinformatics ML.
• SMILE promotes more transparent and powerful applications of ML in biological research.