Preprint
Machine Learning

Machine Learning-Based Virtual Screening and Molecular Simulation Approaches Identified Novel Potential Inhibitors for Cancer Therapy

Muhammad Shahab(State Key Laboratories of Chemical Resources Engineering, Beijing University of Chemical Technology, Beijing 100029, China), Guojun Zheng(State Key Laboratories of Chemical Resources Engineering, Beijing University of Chemical Technology, Beijing 100029, China), Abbas Khan(Department of Bioinformatics and Biological Statistics, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai 200240, China), Dongqing Wei(Department of Bioinformatics and Biological Statistics, School of Life Sciences and Biotechnology, Shanghai Jiao Tong University, Shanghai 200240, China), Alexander S. Novikov(Institute of Chemistry, Saint Petersburg State University, Saint Petersburg 199034, Russia)
August 11, 2023Biomedicines29 citations
DOI

29

Citations

0

Influential Citations

Biomedicines

Venue

2023

Year

Abstract

Cyclin-dependent kinase 2 (CDK2) is a promising target for cancer treatment, developing new effective CDK2 inhibitors is of great significance in anticancer therapy. The involvement of CDK2 in tumorigenesis has been debated, but recent evidence suggests that specifically inhibiting CDK2 could be beneficial in treating certain tumors. This approach remains attractive in the development of anticancer drugs. Several small-molecule inhibitors targeting CDK2 have reached clinical trials, but a selective inhibitor for CDK2 is yet to be discovered. In this study, we conducted machine learning-based drug designing to search for a drug candidate for CDK2. Machine learning models, including k-NN, SVM, RF, and GNB, were created to detect active and inactive inhibitors for a CDK2 drug target. The models were assessed using 10-fold cross-validation to ensure their accuracy and reliability. These methods are highly suitable for classifying compounds as either active or inactive through the virtual screening of extensive compound libraries. Subsequently, machine learning techniques were employed to analyze the test dataset obtained from the zinc database. A total of 25 compounds with 98% accuracy were predicted as active against CDK2. These compounds were docked into CDK2’s active site. Finally, three compounds were selected based on good docking score, and, along with a reference compound, underwent MD simulation. The Gaussian naïve Bayes model yielded superior results compared to other models. The top three hits exhibited enhanced stability and compactness compared to the reference compound. In conclusion, our study provides valuable insights for identifying and refining lead compounds as CDK2 inhibitors.

Analysis

Why This Paper Matters

This paper addresses a critical need in cancer therapy: the discovery of selective CDK2 inhibitors. CDK2 is a validated target, but existing inhibitors often lack selectivity, leading to side effects. The authors combine machine learning with molecular simulation to identify novel lead compounds, showcasing a modern computational approach that can significantly reduce the time and cost of early drug discovery.

The study is particularly relevant as it applies multiple ML classifiers to a real-world drug design problem, demonstrating that even relatively simple models like Gaussian Naïve Bayes can outperform more complex ones when properly trained and validated. This highlights the importance of model selection and evaluation in cheminformatics.

Technical Contributions

  • ML Model Comparison: The study systematically compares k-NN, SVM, RF, and GNB for binary classification of active vs. inactive CDK2 inhibitors, using 10-fold cross-validation to ensure robustness.
  • Virtual Screening Pipeline: A practical pipeline is established: train ML models on known inhibitors, screen large compound libraries (ZINC), and filter by predicted activity.
  • Molecular Docking and MD Simulation: The top ML-predicted hits are validated through docking into the CDK2 active site and further assessed via MD simulations to evaluate stability and compactness.
  • Identification of Novel Hits: Three compounds with promising binding characteristics are proposed as potential leads, with enhanced stability compared to a reference inhibitor.

Results

The Gaussian Naïve Bayes model yielded superior results compared to other models, though specific metrics (e.g., AUC, F1) are not detailed in the abstract. The virtual screening of the ZINC database produced 25 compounds predicted as active with 98% accuracy. After docking, three compounds were selected for MD simulation. These top hits exhibited enhanced stability and compactness relative to the reference compound, suggesting better binding and potential efficacy.

Significance

This research contributes to the growing field of AI-driven drug discovery by providing a validated computational workflow that can be adapted to other targets. The use of ML for virtual screening is not new, but the integration with MD simulation adds a layer of dynamic validation that strengthens the credibility of the hits. The findings offer a starting point for experimental validation and could lead to the development of selective CDK2 inhibitors, addressing an unmet medical need in oncology.

From an AI perspective, the paper underscores the value of classical ML methods in cheminformatics, showing that sophisticated deep learning is not always necessary. It also highlights the importance of rigorous cross-validation and the potential of ML to prioritize compounds for costly experimental assays.

machine-learningvirtual-screeningdrug-discoverycdk2-inhibitorsmolecular-dockingmolecular-dynamicscancer-therapy

Related Papers