Delicious data: Shiru uses machine learning to identify food-friendly proteins

In California, alternative protein startup Shiru has closed a $17 million Series A funding round and says it will use the funds to build a pilot-scale facility for “label-friendly” and animal-free alternatives to food ingredients such as methylcellulose, gelatin, and casein. 

Founder and CEO Dr. Jasmin Hume tells FoodNavigator-USA the company is combining machine learning and fermentation to genetically engineer protein-producing microorganisms. “We mine databases​ of plants​ to identify proteins that have certain functional qualities such as emulsification, gelation, solubility, and foaming that you get in things like dairy proteins, egg proteins, and gelatin,”​ she says. 

Producing proteins in this way is more sustainable than using animals or obtaining ingredients from plants, where the protein might only be present in small amounts. “So we’re developing the first database that connects protein identity with protein food-related function, at scale,” ​added Hume, formerly ​a biochemist at food tech Eat Just. 

The funding was led by S2G Ventures. Shiru hopes to have the pilot facility online early next year.