Last Updated on July 24, 2026 by Staff
Scientists have been using a process called laboratory evolution to create proteins that can help treat diseases. They take a protein that already exists in nature and modify it a bit at a time until it can do something new and useful. This process has led to some breakthroughs but it can also make the proteins unstable.
Now some researchers at the Broad Institute have found a way to do things. They use intelligence to design proteins that are stronger and more stable before they even start modifying them in the laboratory.
They did a study that was published in Nature. It showed that the proteins designed by artificial intelligence worked better than the natural proteins. This could be a deal for developing new medicines and other treatments that use proteins.
Combining AI And Evolution
Laboratory evolution is a tool that scientists use to create new proteins. They make lots of copies of a protein. Then mutate them to see if any of the mutations make the protein better at doing something. They keep doing this until they get a protein that can do what they want.
There is a problem with this process. When proteins get abilities they can also become unstable. This means they do not work well inside living cells.
To fix this problem the researchers used intelligence to redesign the proteins before they started evolving them. They used an AI system called ProteinMPNN to make the proteins more stable.
Making Stronger Proteins
The researchers were working with a type of protein called neurotoxin proteases. These proteins are used in Botox treatments. They can cut specific proteins inside nerve cells.
The scientists wanted to make these proteins cut a protein called ataxin-2 that is associated with some neurodegenerative diseases. They used a technique called PACE to evolve the proteins and make them better at cutting ataxin-2.
The results were really good. The proteins that were designed by intelligence were 79 times better at cutting ataxin-2 than the natural proteins. They stayed stable the whole time.
Better Results
The researchers did the experiment again with proteins and they got the same results. The proteins designed by intelligence always worked better than the natural proteins.
They figured out that this is because the stable proteins can tolerate mutations without becoming unstable. This means they can keep evolving and getting better without breaking down.
The researchers also found out that if they took the mutations from the artificial intelligence proteins and put them into the natural proteins the natural proteins would become unstable. This shows that stability is really important for proteins to evolve.
The Future Of Medicine
This study shows that using intelligence and laboratory evolution together is more powerful than using either one alone. Scientists think this could be used to make all sorts of proteins that are used in medicine, including enzymes for gene editing and treatments for diseases.
This could even make technologies like editing better by making the proteins more stable. The researchers are already working on using this approach with types of proteins.
The big lesson from this study is that starting with a designed protein makes a big difference in what you can achieve with laboratory evolution. As artificial intelligence keeps getting better it could change the way scientists develop medicines and make protein-based treatments safer and more effective.