An AI designed antimicrobial proteins that were then tested in real life and shown to work.
This same approach could eventually be used to create new drugs, something that requires a lot of computation to try various combinations of substances to come up with one that actually works.
In this case, the scientists used an AI called ProGen to design millions of new proteins and then created a small sample to test how well they worked.
ProGen learned how to generate new proteins by learning “the grammar” of the amino acid combination of 280 million existing proteins. Of the 100 physical proteins they created, 66 participated in chemical reactions similar to naturally occurring bacteria-killing proteins in egg whites and saliva, suggesting they can kill bacteria too.
A similar process could be used to create new test molecules for drug development, although they will still have to be tested in laboratories, which is a long and time-consuming process.
The study was conducted by Ali Madani and colleagues at Salesforce Research in California. James Fraser from the University of California San Francisco was also part of the team.
Creating a new drug is a costly and time-consuming process due to several factors, including the rigorous research and development process that often includes animal studies and human clinical trials. There is also a need to be approved by regulatory agencies, but protein blending usually takes the longest. That is why the help of the AI ​​can be essential, since it usually has more aim than that of simple humans.
You have more information on Madani’s page.
