Today, two laboratories separately announced programs to use diffusion models to generate designs for new proteins with more precision than ever before. Generate Biomedicines, a Boston-based startup, revealed a program called Chromawhich the company describes as the “DALL-E 2 of biology.”
At the same time, a team at the University of Washington led by biologist David Baker built a similar program called RoseTTAFold Diffusion. In a preprint paper posted online today, Baker and his colleagues have shown that their model can generate precise designs for new proteins that can then be inserted. life in the laboratory. “We’re making proteins that are really unlike existing proteins,” said Brian Trippe, one of RoseTTAFold’s co-developers.
These protein generators can be directed to create designs for proteins with specific properties, such as shape, size or function. In effect, this helps create new proteins to do specific jobs on demand. The researchers hope that this will eventually lead to the development of new and more effective drugs. “We were able to discover in minutes what went on in evolution over millions of years,” said Gevorg Grigoryan, CEO of Generate Biomedicines.
“What’s remarkable about this work is the creation of proteins under the desired constraints,” said Ava Amini, a physiologist at Microsoft Research in Cambridge, Massachusetts.
Proteins are the basic building blocks of living systems. In animals, they digest food, contract muscles, detect light, control the immune system, etc. When people are sick, protein plays an important role.
Therefore, protein is the main target of the drug. And many of today’s newest drugs are protein-based. “Nature uses protein for everything,” says Grigoryan. “The promise of therapeutic interventions is really huge.”
But drug designers now have to draw up an ingredient list made up of natural proteins. The goal of protein creation is to expand that list to an almost infinite group of computer-designed proteins.
Computational techniques for designing proteins are not new. But previous approaches have been slow and ineffective in designing large proteins or protein complexes—molecular machines made up of many proteins linked together. And such proteins are often important for treating disease.