Pharma consortium forms to train AI on 10,000 antibodies

Ginkgo Datapoints and Berlin's Apheris have launched a consortium with AbbVie, argenx, Lundbeck, and Takeda to standardize antibody data for drug discovery AI. Each company shares proprietary antibody info using Apheris' privacy-preserving system, aiming for a combined dataset of 10,000 antibodies. The goal is to improve how AI spots which antibody drugs could be made and brought to patients, overcoming the limits of small, scattered datasets.

• Consortium unifies effort from four major pharma firms
• Aims for a combined dataset of 10,000 antibodies
• Members use Apheris infrastructure, keeping data private
• AI models are benchmarked on the full consortium dataset
• Ginkgo oversees sequence selection and model foundation