AlzyFinder Platform: A web-based tool for machine-learning based virtual screening and network pharmacology

Alzheimer’s disease (AD), a prevalent neurodegenerative disorder, presents significant challenges in drug development due to its multifactorial nature. The AlzyFinder Platform presented here addresses this by providing a comprehensive, free web-based tool for ligand-based virtual screening and network pharmacology, specifically targeting over 85 key proteins implicated in AD. Utilizing advanced machine learning models, AlzyFinder facilitates the identification of potential multitarget ligands and their systemic therapeutic impacts. The platform's user-friendly interface allows for various input methods, including SMILES format and a molecular editor. AlzyFinder outputs interaction data in multiple formats (tables, heatmaps, and interactive Ligand-Protein Interaction networks) enhancing the visualization and analysis of drug-protein interactions. Key machine learning models employed by AlzyFinder, were meticulously trained and validated using robust methodologies, ensuring high predictive accuracy. Machine Learning models were built with XGBoost algorithm, optimized through Optuna and evaluated based on balanced accuracy, precision, and F1 score metrics. A unique soft-voting ensemble approach further refines the classification process, integrating the strengths of individual models. Validation included extensive testing with active, inactive, and decoy molecules, demonstrating the platform’s efficacy in distinguishing active compounds. AlzyFinder’s innovative approach extends beyond traditional virtual screening by incorporating network pharmacology analysis, thus providing insights into the systemic actions of drug candidates. This feature allows for the exploration of ligand-protein and protein-protein interactions and their extensions, offering a comprehensive view of potential therapeutic impacts. As the first open-access platform of its kind, AlzyFinder stands as a valuable resource for the AD research community, available at http://www.alzyfinder-platform.udec.cl with supporting data and scripts accessible via GitHub https://github.com/ramirezlab/AlzyFinder. In this work as case of study we screened 5 molecules recently reported as active compounds against five key AD targets. We observe that AlzyFinder was able to accurately predict (with a probability greater than 0.70) all five molecules as true positives.