Salesforce scientists develop proteins with AI that can eat garbage


Header image for this article

What do you get when the world’s largest CRM research enters the industry and leverages AI to build its products? You understand Progen, a new AI system that can create synthetic enzymes from scratch that can act like real ones found in nature. ProGen was created by Salesforce Research (yes, that Salesforce) and uses language processing to learn about biology. In short, progen takes the sequence of amino acids and turns them into proteins.

In 1999, biologist Günter Blobel Won the Nobel Prize for its work in protein synthesis, but this new AI-powered technology is already going beyond that. ProGen accelerates the creation of new proteins, which can be used for many things such as medicine or to break down plastic in landfills, perhaps helping us avoid the great garbage avalanche of 2505. .

“Synthetic designs are better than conventional designs,” said James Fraser, a scientist involved in the project. “We can now make specific types of enzymes, such as those that work well in hot temperatures or acids.”

To build ProGen, Salesforce scientists fed the system amino acid sequences from 280 million different proteins. The AI ​​system quickly generated an astonishing 1 million protein sequences, of which 100 were selected for testing. Five of these were cloned into actual proteins and tested in cells. This is only 0.0005% of the results produced! The next frontier seems to be developing AI to test all possibilities. The two synthetic enzymes were as good at breaking down bacteria as the natural enzymes found in egg whites. Even still, the two were only 18% similar.

ProGen was built in 2020 using LLM which was originally built to text like ChatGPT. The AI ​​system learned the principles and structure of proteins by looking at lots of data. With proteins, there is a huge number of possibilities, but ProGen can still make working enzymes, even when the results vary greatly.

“This is a new tool for protein engineers and we’re excited to see what it can be used for,” said Ali Madani, a scientist involved in the project. The project seems incredibly expensive, and would have cost Salesforce a fortune to move forward, so we were surprised to see that ProGen’s code Available on Github. For anyone who wants to try it (or add to it).

Source: Sales force through The new scientist


You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *