Predicting the efficiency of chemical compounds against the Tuberculosis bacterium
What is it about?
In recent years there have been screens of millions of compounds yielding over 1500 inhibitors of Mycobacterium tuberculosis (Mtb). We demonstrate learning from in vivo active and inactive compounds using machine learning classification models consisting of 773 compounds.
Why is it important?
A Bayesian model predicts 8 out of 11 additional in vivo actives not included in the model as an external test set. Curation of 70 years of Mtb data can provide statistically robust computational models to focus resources on in vivo active small molecule antituberculars.
The following have contributed to this page: Dr Antony John Williams, Dr Sean Ekins, Joel Freundlich, and Alex Michael Clark