The same technology that powers DALL-E and other image-generating AIs could one day be used by researchers to design antibodies to treat things like cancer and infectious diseases.
Challenge: Your body makes antibodies that attach to unwanted substances such as viruses and bacteria so your immune system can remove them.
Decades ago, researchers discovered that antibodies could be created in the lab. The proteins can then be grown and used as medicines to help the body recognize and attack pathogens, cancer cells, toxic substances, and more.
“Ten years from now, this is how we will be designing antibodies.”
Nathaniel Bennett
However, existing methods for designing therapeutic antibodies are not very good.
Once researchers have a target for antibody therapy, such as a specific type of cancer cell, they can inject it into animals to see what types of antibodies their immune systems produce. The most promising ones can then be selected and purified for use in humans.
Alternatively, you could screen large databases of known antibodies for antibodies likely to bind to your target and test them to fine-tune the best option, but both processes are difficult, expensive, and time-consuming. It takes.
New method: Researchers at the University of Washington (UW) have developed AI for antibody design. They shared details of the system, which is currently in the proof-of-concept stage, on the preprint server bioRXiv.
“Ten years from now, this is how we will be designing antibodies,” study co-author Nathaniel Bennett told Nature.
diffusion: In 2023, a team at Wisconsin State University announced RFdiffusion, a tool that uses diffusion models (a type of technique also used in image generation AI) to generate new protein designs. The new study improved the tool by training it on images of antibodies bound to their targets.
They then used AI to design antibodies that they predicted would bind to proteins on a variety of targets, including viruses, bacteria, and cancer cells. Finally, we actually created some antibodies in the lab and saw how well they attached to their targets.
“This feels like a very groundbreaking moment. It shows that it's actually possible.”
joseph watson
Future prospects: The success rate was modest, about 1%, and the antibodies that bound to the target did not form particularly strong bonds. Still, the researchers believe their approach can save them time and money, and now they know it. can While you're working, you can focus on improving.
“This is a proof-of-principle study,” study co-lead Joseph Watson told Nature. “This feels like a very groundbreaking moment. It shows that it's actually possible.”
We look forward to hearing from you! If you have comments about this article or tips for future Freethink stories, please send an email to: tips@freethink.com.
