Scientists are taking advantage of generative artificial intelligence (AI) to make antibodies more effective against viruses like SARS-CoV-2 and ebola. Stanford University biochemist Peter Kim shares that AI-generated antibody designs have access to information otherwise not obtainable by experts. This is possible using neural networks that can generate content based on previously established patterns.
Immune system proteins, known as antibodies, have become a staple in biotechnology and this new AI technology has the potential to open the door to treating targets otherwise unresponsive to the existing approaches. Meta AI researchers in New York City, a subsidiary of tech giant Meta, created a protein language model to identify limited antibody mutations. Surprisingly, these models further improved the capacity of the antibodies to adhere to their respective targets for viruses like SARS-CoV-2, ebola, and influenza.
A challenge when designing completely new antibodies is related to the flexibility of their loops, and AI processes can be effective in modeling their corresponding interactions. At the same time, AI is also able to design more accurately target-specific antibodies as well as drugs that can dual-target proteins such as a tumor and its respective immune cell destroyer.
Meta is a high-tech analytics services firm headquartered in New York City and founded in 2000. They offer services across areas such as AI, cloud, digital marketing, eCommerce, and salesforce. They have created the Meta AI research subsidiary to develop the protein language model used in the research of antibody design.
Peter Kim is a biochemist at Stanford University. He became interested in the development of large scale application of AI to biochemistry and antibody engineering. He has since worked to establish use cases for AI and its advances in antibody design.
