Integrated Metabolomics-KPCA-Machine Learning framework: a solution for geographical traceability of Chinese Jujube.

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Data Science

Background:

• Chinese jujube (CJ) faces significant challenges in geographical traceability due to widespread product adulteration.
• Ensuring the authenticity of agricultural products is crucial for global trade and consumer trust.

Purpose of the Study:

• To develop a robust origin identification system for Chinese jujube using advanced analytical techniques.
• To establish a reliable method for tracing the geographical origin of agricultural products.

Main Methods:

• Untargeted metabolomics using LC-MS/MS to identify metabolites in CJ samples.
• Multivariate analysis to identify key discriminant variables.
• Kernel Principal Component Analysis (KPCA) for dimensionality reduction.
• Machine learning (ML) algorithms, including K-means clustering, for classification.

Main Results:

• Identified 312 metabolites, with 37 key discriminant variables.
• KPCA effectively compressed features into 28 principal components, retaining 90.59% of the information.
• The Metabolomics-KPCA-ML framework significantly improved sample differentiation and achieved accurate classification of CJ origins.

Conclusions:

• The developed "Metabolomics-KPCA-ML" paradigm offers a powerful solution for the geographical traceability of Chinese jujube.
• This approach provides a reliable method for authenticating geographical indication agricultural products.
• The study demonstrates the potential of integrating metabolomics and machine learning for agricultural product traceability.