Accelerating Drug Discovery with Artificial Intelligence: EquiBind Finds Potential Drug Molecules Thousand Times Faster

Model of artificial intelligence finds drug molecules up to a thousand-fold faster

There are an infinite number molecules in the universe. What fraction of molecules has the potential to be drug-like and can be used for developing life-saving drugs? Millions? Billions? Trillions? Answer: novemdecillion or 1060. This staggering number is beyond the capabilities of existing drug design models and prolongs drug development for diseases that are spreading rapidly, like COVID-19. For perspective, the Milky Way contains about 100 million stars, or 108.

In a paper to be presented at the International Conference on Machine Learning, MIT researchers have developed a geometric deep learning model called EquiBind. This model is 1,200-times faster than QuickVina2-W in binding drug-like molecules with proteins. EquiBind was developed from EquiDock. EquiDock specialized in binding two protein using a technique created by the late Octavian Eugen Ganea.

Drug discovery is a process that drug researchers use to find promising molecules similar to drugs which can \”dock\” or bind properly onto specific protein targets. The binding drug (also known as ligand) can prevent a protein’s function after docking successfully to the protein. When this happens to a protein essential to a bacterium it can kill that bacterium and confer protection to the body.

Source:
https://techxplore.com/news/2022-07-artificial-intelligence-potential-drug-molecules.html