News Artificial Intelligence

The AI model OpenFold3 takes a crucial step in making protein predictions

Scientists open up AlphaFold3’s secrets with the new OpenFold3 model

Ribbon diagrams of OpenFold3’s predictions of proteins compared with experimentally determined structures.

An AI model called OpenFold3 is able to predict how proteins interact with a variety of other molecules. Shown are ribbon diagrams of OpenFold3’s predictions of proteins (blue) compared with experimentally determined structures (gray). At left, the human PDE10 protein, a target for potential new schizophrenia drugs, has an inhibitory molecule nestled within. The interferon regulatory factor 4, a multiple myeloma drug target, docks with DNA (top right) and an antibody grabs another protein (bottom right).

Lukas Jarosch/Columbia University

By Tina Hesman Saey

A new AI model is opening the black box of the leading artificial intelligence tool for predicting how proteins will interact with small molecules, such as drugs.