Related Experiment Video
Updated: Aug 5, 2026
10:29
Quantitative Structure-Activity Relationship, Activity Prediction, and Molecular Dynamics of Non-nucleotide Reverse Transcriptase Inhibitors
Published on: May 9, 2025
Explainable Artificial Intelligence (XAI) and Molecular Modeling Techniques to Discover Putative HER2 Inhibitors
Shailima Rampogu1, Thananjeyan Balasubramaniyam2, Cheol-Hee Yoon3
1Cachet Big Data Lab, Hyderabad 500081, Telangana, India.
International Journal of Molecular Sciences
|July 28, 2026
Summary
Two marine natural products, CMNPD30448 and CMNPD7060, show potential as HER2 inhibitors for breast cancer treatment. Computational models predicted their activity and stability, warranting further investigation.
Area of Science:
- Pharmacology
- Computational Chemistry
- Oncology
Background:
- Breast cancer remains a leading cause of mortality in women worldwide.
- The human epidermal growth factor receptor 2 (HER2) is a validated therapeutic target in breast cancer treatment.
- Marine natural products offer a rich source of novel drug candidates.
Purpose of the Study:
- To identify potential HER2 inhibitors from a marine natural product database using structure-based pharmacophore modeling.
- To computationally validate the identified compounds through molecular dynamics simulations and machine learning models.
- To explore the key molecular features contributing to the inhibitory potential of the candidate compounds.
Main Methods:
- Structure-based pharmacophore model generation and screening of the Comprehensive Marine Natural Product Database (CMNPD).
- Molecular dynamics simulations (500 ns) to assess the stability of top-scoring compounds.
- Machine learning (ML) and neural network (NN) models, including random forest classification with PubChem fingerprints, were employed on ChEMBL compounds.
- Explainable AI (LIME) was used to interpret the ML model's predictions.
Main Results:
- Two compounds, CMNPD30448 (hit1) and CMNPD7060 (hit2), exhibited superior docking scores compared to the co-crystallized ligand.
- Molecular dynamics simulations revealed stable Root Mean Square Deviation (RMSD), Radius of Gyration (Rg), and Root Mean Square Fluctuation (RMSF) for both hits.
- The random forest model achieved high accuracy (0.91) and ROC-AUC (0.96), predicting the retrieved compounds as active HER2 inhibitors.
- LIME analysis identified specific PubChem fingerprints contributing to the inhibitory activity of CMNPD30448 and CMNPD7060.
Conclusions:
- CMNPD30448 and CMNPD7060 are identified as promising HER2 inhibitors based on in silico analyses.
- The computational approach provides a strong foundation for further experimental validation.
- These marine-derived compounds represent potential leads for developing novel breast cancer therapeutics.
