Classification of multidrug-resistance reversal agents using structure-based descriptors and linear discriminant...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Multidrug resistance (MDR) is a major challenge in cancer chemotherapy.
• Identifying effective MDR reversal agents (MDRRAs) is crucial for improving treatment outcomes.
• Existing screening methods can be time-consuming and costly.

Purpose of the Study:

• To develop and validate computational models for classifying potential MDRR agents.
• To utilize structure-based descriptors for predicting compound activity.
• To establish a rapid screening mechanism for identifying novel MDRRAs.

Main Methods:

• Linear discriminant analysis (LDA) was employed to build classification models.
• Structure-based topological descriptors were used to encode molecular features.
• Models were trained and evaluated using activity data from 609 compounds against adriamycin-resistant P388 murine leukemia cells.
• Two classification schemes (three-class and two-class problems) and two activity distributions were considered.

Main Results:

• A nine-topological-descriptor model achieved 83.1% correct classification for inactive/moderately active/active compounds with small activity separation.
• A six-topological-descriptor model reached 92.0% correct classification with larger activity separation.
• Monte Carlo cross-validation confirmed the robustness of the developed models.

Conclusions:

• The developed LDA models effectively classify MDRR agents based on their activity.
• These models demonstrate potential as a screening tool for identifying novel MDRR agents from large compound libraries.
• The study highlights the utility of structure-based descriptors in computational drug discovery for MDR.