How Salesforce’s AI-Designed Proteins Could Help Uncover Potential Medical Treatments

Salesforce ProGen — an AI language model trained on the largest protein database available — was featured in this month’s edition of Nature Biotech to show how generative AI can lead to potential solutions for addressing challenges in human disease and the environment.

The Nature Biotech story also reveals the first known 3D structure of an artificial protein designed fully by AI.

Why it’s important: Salesforce ProGen was built on the same fundamental idea as the large-language chatbots that have recently entered the zeitgeist. 

• The unique element? ProGen gives scientists and researchers the ability to use AI as a controllable tool to design highly-tailored proteins with desired properties. 
• In the future, these properties can include the ability to bind to another molecule and operate at high temperatures.

In fact, the Salesforce AI Research team is currently leveraging ProGen to identify potential treatments for neurological and autoimmune disorders such as rheumatoid arthritis and multiple sclerosis. 

Fast fact: Using ProGen, the Salesforce AI Research team tested artificial proteins against natural proteins for antibacterial function in a lab, working with Tierra Biosciences and Professor James Fraser’s lab at the University of California San Francisco. 

• Results showed that 73% of ProGen’s artificially generated proteins were functional, compared to 59% of natural proteins.

The Salesforce perspective: “ProGen exemplifies how generative AI can be applied to bioscience and healthcare in this emerging field,” said Nikhil Naik, Director of Research, Salesforce. “This research reveals how useful large language models can be in designing novel proteins for use in real-world applications.”

ProGen exemplifies how generative AI can be applied to bioscience and healthcare in this emerging field.

Nikhil Naik, Director of Research, Salesforce

The big picture: As the world continues to combat disease, evolving proteins from scratch in a laboratory has traditionally been a slow process. 

• With AI models like ProGen, scientists can help develop a proactive approach for designing proteins. 
• In the future, this could help quickly develop treatments for diseases and enzymes that could have industrial and environmental applications.

What’s next: Explore more about ProGen here, and find more Salesforce AI Research insights on the team’s AI Research blog. 

• And, read how Salesforce aims to build products that are ethical by design.