Multilabeled classification approach to find a plant source for terpenoids.

Area of Science:

• Chemotaxonomy
• Computational Chemistry
• Plant Sciences

Background:

• Sesquiterpene lactones (STLs) are secondary metabolites found in the Asteraceae family.
• Previous models assigned STLs to a single plant tribe, but STLs often occur in multiple tribes.
• This limitation necessitates a more sophisticated classification approach.

Purpose of the Study:

• To develop and evaluate a multilabel classification model for STLs within the Asteraceae family.
• To explore the utility of multilabel classification in understanding STL distribution across plant tribes.
• To improve the efficiency of discovering STLs from specific plant sources.

Main Methods:

• Overview of techniques for examining multilabeled data.
• Discussion on evaluating multilabeled classifier performance.
• Application of cross-training with support vector machines (ct-SVM) and multilabeled k-nearest neighbors (ML-kNN) to STL classification into seven Asteraceae tribes.

Main Results:

• The multilabel approach more accurately reflects the natural occurrence of STLs across multiple tribes.
• Improved understanding of the chemotaxonomic relationships between different Asteraceae tribes based on STL profiles.
• A significant reduction in the number of plant sources needing investigation to find specific STLs.

Conclusions:

• Multilabel classification provides a more realistic model for STL distribution in Asteraceae.
• This approach enhances chemotaxonomic insights and aids in the targeted discovery of bioactive natural compounds.
• The developed models support efficient plant collection for identifying STLs with desired properties.