BioMarin has turned to an AI lab to widen its pipeline of enzyme therapies. The company will work with Profluent, an Emeryville startup, on designing and optimizing protein drugs for genetically defined conditions.
Profluent’s models are trained on a sequence library it calls the Profluent Protein Atlas. The company says that collection holds more than 115 billion unique entries. BioMarin will screen the generated designs for the activity and stability it needs, and the strongest candidates are earmarked for preclinical work.
The appeal is speed. Therapeutic enzymes are notoriously difficult to engineer, and traditional directed evolution can swallow years of bench work before a candidate is ready to test.
Pooja Agarwal, who leads metabolic conditions research at BioMarin, said the collaboration adds outside innovation that can complement internal work and move science forward faster. Profluent chief executive Ali Madani said his company’s latest foundation models were a fit for BioMarin’s record of reaching patients.
Profluent was founded in 2022 and counts Altimeter Capital, Bezos Expeditions and Spark Capital among its backers.
