Related Experiment Video
Updated: Nov 30, 2025
06:08
Author Spotlight: A Pseudotype Virus System for Assessing Omicron Subvariants and Neutralizing Antibodies in SARS-CoV-2 Research
Published on: September 8, 2023
1.5K
REDIAL-2020: A suite of machine learning models to estimate Anti-SARS-CoV-2 activities
K C Govinda1,2, Giovanni Bocci3, Srijan Verma2,4
1Computational Science Program, The University of Texas at El Paso, Texas 79968, USA.
Chemrxiv : the Preprint Server for Chemistry
|November 17, 2020
Summary
REDIAL-2020 is a machine learning tool that predicts small molecule activity against SARS-CoV-2. It aids drug discovery for COVID-19 by estimating antiviral potential from molecular structure.
Area of Science:
- Computational Chemistry
- Drug Discovery
- Machine Learning
Background:
- Urgent need for drug discovery and repositioning strategies for COVID-19.
- High-throughput screening data from the NCATS COVID19 portal is crucial for developing predictive models.
Approach:
- Developed REDIAL-2020, a suite of machine learning models using fingerprint, physicochemical, and pharmacophore descriptors.
- Trained models on SARS-CoV-2 assay data using various machine learning algorithms.
- Created an ensemble consensus predictor combining multiple models for improved performance.
Key Points:
- REDIAL-2020 estimates anti-SARS-CoV-2 activity from molecular structure.
- The web application predicts activity across viral entry, replication, and infectivity.
- Models demonstrate 60-74% external predictivity on independent datasets.
Conclusions:
- REDIAL-2020 serves as a rapid online tool for identifying potential COVID-19 therapeutics.
- The models and source code are publicly available via DrugCentral, GitHub, and Docker Hub.
