PLASMOpred: A Machine Learning-Based Web Application for Predicting Antimalarial Small Molecules Targeting the Apical...

Area of Science:

• Computational chemistry and parasitology
• Drug discovery and development
• Machine learning applications in medicine

Background:

• Falciparum malaria remains a significant global health threat, causing over half a million deaths annually.
• Parasite invasion of human erythrocytes is essential for malaria survival, involving key proteins like Apical Membrane Antigen 1 (AMA-1) and Rhoptry Neck Protein 2 (RON2).
• Inhibiting the AMA-1-RON2 interaction blocks parasite invasion, presenting a promising strategy for antimalarial drug development.

Purpose of the Study:

• To leverage machine learning (ML) to predict small molecule inhibitors targeting the AMA-1-RON2 interaction.
• To identify novel antimalarial compounds for further chemotherapeutic exploration.
• To develop a predictive tool for antimalarial drug discovery.

Main Methods:

• Utilized a dataset of 364,447 inhibitors and non-inhibitors from PubChem (AID 720542).
• Employed Morgan fingerprints for feature extraction and balanced data using Synthetic Minority Oversampling Technique.
• Developed and evaluated five ML models: Random Forest, Gradient Boost Machines, CatBoost, AdaBoost, and Support Vector Machine, assessing performance via accuracy, F1 score, and ROC-AUC.

Main Results:

• Gradient Boost Machines (GBMs) demonstrated superior performance with 89% accuracy and 92% ROC-AUC.
• CatBoost (CB) and Random Forest (RF) models also showed high efficacy, with ROC-AUC scores of 93% and 91%, respectively.
• Applicability domain analysis confirmed the reliability of predictions within a defined Tanimoto distance threshold.

Conclusions:

• Experimentally validated AMA-1-RON2 inhibitors are valuable starting points for next-generation antimalarial drugs.
• The developed ML models, deployed as the web application PLASMOpred, offer a powerful tool for identifying potential antimalarial compounds.
• This study highlights the potential of machine learning in accelerating the discovery of effective treatments for falciparum malaria.