Machine Learning, Molecular Docking, and Dynamics-Based Computational Identification of Potential Inhibitors against...

Area of Science:

• Computational chemistry and pharmacology
• Drug discovery and development
• Machine learning in bioinformatics

Background:

• Lung cancer remains a leading cause of cancer mortality globally.
• Acquired resistance to existing therapies presents a major clinical challenge.
• Phytomolecules offer diverse chemical structures for novel drug development.

Purpose of the Study:

• To develop a machine learning model for identifying phytomolecule-based lung cancer inhibitors.
• To screen a large natural product database for potential anti-lung cancer agents targeting EGFR mutants.
• To validate promising natural product candidates using molecular docking and dynamics simulations.

Main Methods:

• Development and comparison of four machine learning models (k-NN, RF, SVM, XGBoost) with MACCS and Morgan2 fingerprints.
• Virtual screening of ~400,000 natural products from the COCONUT database using docking against EGFR mutants.
• Molecular dynamics simulations and structural similarity analysis of top-ranked indolocarbazole compounds.

Main Results:

• The random forest model with MACCS fingerprints demonstrated superior performance in predicting inhibitory activity.
• A multistep screening identified 205 potential natural product inhibitors, with a notable enrichment of indolocarbazole scaffolds.
• Top indolocarbazole molecules exhibited strong binding affinity and stability via molecular dynamics, showing similarity to known EGFR mutant inhibitors.

Conclusions:

• Machine learning, molecular docking, and dynamics simulations effectively identified promising natural product inhibitors for lung cancer.
• Indolocarbazole-based natural products represent a highly promising class of compounds for targeting lung cancer, particularly EGFR mutants.
• This study highlights the potential of integrating computational approaches for accelerating the discovery of novel anti-cancer therapeutics.