Related Experiment Video
Updated: Mar 26, 2026
10:11
Modeling The Lifecycle Of Ebola Virus Under Biosafety Level 2 Conditions With Virus-like Particles Containing Tetracistronic Minigenomes
Published on: September 27, 2014
37.2K
Machine learning models identify molecules active against the Ebola virus in vitro
Sean Ekins1,2,3, Joel S Freundlich4, Alex M Clark5
1Collaborations in Chemistry, Fuquay-Varina, NC, 27526, USA.
F1000Research
|August 8, 2017
Summary
Machine learning models identified novel Ebola virus (EBOV) inhibitors from existing drugs. Validated models prioritized compounds, leading to the discovery of potential EBOV treatments like quinacrine, pyronaridine, and tilorone.
Area of Science:
- Virology
- Computational Biology
- Drug Discovery
Background:
- Ebola virus (EBOV) infection remains a significant global health threat.
- High-throughput screening (HTS) has identified FDA-approved drugs with anti-EBOV activity.
- Millions of compounds are available for screening, necessitating efficient prioritization methods.
Purpose of the Study:
- To develop and validate machine learning models for predicting EBOV inhibitors.
- To virtually screen compound libraries using these models to identify potential drug candidates.
- To computationally prioritize existing drugs for anti-EBOV activity.
Main Methods:
- Generated Bayesian machine learning models using viral pseudotype entry and EBOV replication assay data.
- Validated the predictive models both internally and externally.
- Computationally scored the MicroSource library of drugs to identify potential inhibitors.
Main Results:
- Three top-scoring compounds—quinacrine, pyronaridine, and tilorone—showed significant in vitro anti-EBOV activity (EC50 values of 350, 420, and 230 nM, respectively).
- Pyronaridine, a component of an EMA-approved antimalarial, shares a scaffold with known EBOV-active drugs.
- Tilorone and quinacrine also demonstrated antiviral potential, with tilorone exhibiting broad biological activities.
Conclusions:
- Validated machine learning models can be generated from datasets with fewer than 1,000 molecules.
- These models are effective in identifying novel Ebola virus inhibitors.
- Computational screening and machine learning offer a viable strategy for prioritizing compounds for drug discovery.
