Artificial intelligence simulates drug discovery

Artificial intelligence simulates drug discovery

Their are 10^60 molecules in the universe that can be used to develop life-saving drug treatments.

A geometric deep-learning model called EquiBind is developed that is 1200 times faster than older computational methods.

EquiBind is based on EquiDock, which was invented by the late Octavian-Eugen Ganea and specialised in binding two proteins.

Drug discovery is the technique that drug researchers use to identify possible drug-like compounds that can attach or "dock" correctly onto certain protein targets before drug development ever begins.

The binding drug, also known as the ligand, can prevent a protein from functioning once it has effectively docked to the protein.

When this occurs to a bacteria's vital protein, the bacterium can be killed, providing protection for the human body.

Heavy candidate sampling along with techniques like scoring, ranking, and fine-tuning to acquire the best "match" between the ligand and the protein. It is costly too.

Computational methods includes

It is characterised as "blind docking" also simulates drug discovery when EquiBind directly predicts the specific key position in a single step without knowing the protein's target pocket beforehand. 

Deep learning methods includes

EquiBind was successful in binding the ligands that were effective on those proteins, but the majority of conventional docking techniques were unsuccessful.

Other Stories