Identifying Significant Metabolic Pathways Using Multi-Block Partial Least-Squares Analysis.

Area of Science:

• Metabolomics
• Systems Biology
• Bioinformatics

Background:

• Identifying altered metabolic pathways is key to understanding disease mechanisms.
• Current pathway analysis methods often overlook interactions between pathways.
• Existing methods require pre-identification of significant metabolites or model pathways separately.

Purpose of the Study:

• To develop a novel metabolic pathway identification method using multi-block partial least squares (MB-PLS).
• To integrate all metabolic pathways into a global model for enhanced biological interpretation.
• To introduce a new metric, pathway importance in projection (PIP), for pathway significance evaluation.

Main Methods:

• Metabolites were assigned to pathway blocks based on KEGG database annotations.
• Metabolite data matrices were reconstructed into data blocks.
• A multi-block partial least squares (MB-PLS) model was constructed.
• Pathway significance was assessed using the proposed pathway importance in projection (PIP) metric.

Main Results:

• The proposed MB-PLS method demonstrated effectiveness in identifying significant metabolic pathways.
• Results from a simulated dataset showed high consistency with pre-defined perturbed pathways.
• Analysis of a real metabolomics dataset yielded significant pathways consistent with published literature.
• The pathway importance in projection (PIP) metric proved valuable for evaluating pathway significance.

Conclusions:

• The novel MB-PLS-based method provides an effective approach for metabolic pathway identification.
• This method facilitates a more comprehensive biological interpretation of metabolomics data by considering pathway interactions.
• The pathway importance in projection (PIP) metric aids in pinpointing crucial pathways altered by biological perturbations.