AI-fueled protein design startup spun out from Cyrus Biotechnology
Cyrus and Levitate Bio deploy AI-fuelled algorithms to simulate, analyze and design proteins. Their offerings are based on the Rosetta biomolecular modeling suite, which has its roots in the laboratory of David Baker at the University of Washington, US and was developed by a multi-institution collaboration called the Rosetta Commons.
Cyrus was spun out from the Baker lab in 2014 to develop an intuitive online version of Rosetta called Cyrus Bench, which has been sold to over 120 pharmaceutical firms globally. In 2019, Cyrus switched focus to using the platform to develop biologic drugs, with its lead candidate in preclinical development for the treatment of autoimmune conditions.
Wholly owned by the Rosetta Commons Foundation, Levitate Bio will take on the software management, services, and IT operations side of the business, allowing Cyrus to increase its focus on drug development as a customer of Levitate. Other industries that could benefit from Levitate Bio’s AI-based software include agriculture, synthetic biology, manufacturing, and research.
“The AI algorithms for protein design have reached a point where we can create almost any protein, but they are still underutilized in biopharma companies. I am excited for Levitate to bring these tools to as many scientists as possible, so they can use them to discover new drugs,” Baker said in a public statement.
Levitate Bio was founded amid a trend in the software industry of nonprofit organizations being paired with a for-profit business, Sam DeLuca, CEO of Levitate Bio, told Biopharma Reporter in an email. The Mozilla Foundation and the Mozilla Corporation, developer of the Firefox internet browser, are a direct inspiration for the model of Levitate Bio and the Rosetta Commons Foundation, he said.
AI gains momentum in protein engineering
AI-powered tools have a vast potential for making protein design and modeling faster and cheaper than traditional methods, but deploying them in a commercial biotechnology setting requires user know-how and “substantial hardware resources, along with IT and cluster operations skills that many biotech companies lack,” DeLuca said.
There are several competitors in the space that aim to address these problems with deploying AI tools in protein design, including 310, Cradle and Schrodinger. Levitate stands out with its customer service, product design expertise, and adaptability to an individual company’s needs, DeLuca said.
Given the massive progress of AI tools over the past five years, DeLuca sees the adoption of AI in protein engineering transforming the field into a true engineering discipline over the next decade. However, he cautioned that there are still many gaps in our understanding of protein folding.
“We will be limited more by our understanding of biology than by our ability to engineer new therapeutics. While drug discovery will remain an immensely challenging field for the foreseeable future, these new methods are a huge step forward,” he said.
