From Pixels to Proteins: How AI is Changing Protein Creation

Proteins are the foundation of all life on earth.

A team of U.S. researchers has created an artificial intelligence program that can design custom proteins that may speed efforts to design everything from cancer-fighting medicines to proteins that can suck carbon out of the atmosphere.

This program is called RFdiffusion. When asked to create possible drugs, it begins by analyzing a 3D model of a target protein, then designs completely new proteins to bind to the target. The program starts off with shots in the dark as to the protein’s structure, then quickly makes tweaks to the architecture, which is calculated to bind ever more tightly to the target protein. Compared with other non-AI protein design programs, the authors note, RFdiffusion improved the success rate of finding tight protein binds dramatically.

As mentioned, RFdiffusion could have potentially endless uses. By synthesising proteins evolution hasn’t ever produced, it would be of great use creating biomaterials and medicine.

The tools are inspired by AI software that synthesizes realistic images, such as Midjourney. A similar approach to that, researchers have found, can churn out realistic protein shapes to criteria that designers specify— which implies, for example, that it’s possible to speedily draw up new proteins that should bind tightly to another biomolecule. And early experiments show that when researchers manufacture these proteins, a useful fraction do perform as the software suggests.

However, there may be limitations. Currenty, the program has aninability to create proteins that are vastly different from natural proteins, That is because the AI image systems have been trained only on existing proteins that scientists have characterized, he says, and the program tends to create proteins that resemble those. Generating more-alien-looking proteins might require a better understanding of the physics that imbues proteins with their function, which would require more training of the AI.

We’ll have to see where this new technology leads us.