Viral infections, especially those affecting the respiratory system, represent a significant and growing global health concern. Using machine learning (ML) models, we focus on curating comertial datasets of compounds and predicting their antiviral efficacy against respiratory infections. The study specifically addresses seven key respiratory targets associated with viral diseases, leveraging ML classification models trained on a curated dataset of compounds with known antiviral biological activity. These models are used to predict and classify compounds, ultimately facilitating the design of novel antiviral libraries.
- Design antiviral compound libraries targeting respiratory viruses using machine learning.
- Leverage ML classification models to evaluate and predict the antiviral activity of compounds.
- Curate and train models based on biological activity data associated with respiratory diseases.
- Classify compounds into antiviral-focused libraries tailored to specific respiratory targets.
- Provide a reliable resource for researchers to identify potential antiviral candidates against respiratory viruses.
- ML Classification Models: The models used in this study predict the antiviral activity of compounds based on their chemical structures and biological activity.
- Targeted Libraries: The study focuses on seven respiratory targets related to viral infections.
- Data Curation: The models were trained on a curated dataset from a major public compound database annotated with antiviral biological activity data.
- High-Impact Relevance: This approach aims to contribute directly to the urgent need for effective antiviral therapies, especially in the context of respiratory infections like SARS-CoV-2.
-
Data Collection and Curation: Compounds associated with antiviral activity for respiratory diseases were sourced from ChEMBL database. These compounds were annotated with known biological activity.
-
Machine Learning Model Training: The annotated dataset was used to train ML classification models that predict the antiviral activity of compounds. These models were validated and retrained to enhance performance and accuracy.
-
Antiviral Library Design: Once trained, the models were applied to design focused libraries of compounds that specifically target viral respiratory infections. The libraries are designed to be tested against seven key respiratory targets.
-
Validation and Testing: The models' predictions were validated using biological assays and experimental data to ensure their relevance and reliability.
Valle-Núñez G., Cedillo-González R., Avellaneda-Tamayo JF., Saldívar-González FI., Prado-Romero DL., Medina-Franco JL. Machine learning-driven antiviral libraries targeting respiratory viruses. Digital Discovery. 2025 (sent)
This project is licensed under the MIT License - see the LICENSE file for details.