Tiny changes to a therapeutic protein can enhance its performance or initiate a dangerous immune reaction. Novo Nordisk knows that all too well after dropping a Phase III drug trial in 2012 for its engineered Factor VII blood-clotting protein designed to treat hemophilia. Using publicly available software, Novo Nordisk and Food & Drug Administration researchers began outlining methods to predict how immune cells chop up and display engineered proteins on their surface. Their model has shown that two out of three engineered changes in the Factor VII protein activate unwanted T cell immune responses in some people (Sci. Transl. Med. 2017, DOI: 10.1126/scitranslmed.aag1286). One determining factor for the response is a person’s unique HLA genes, which are involved in presenting foreign protein fragments to T cells. The researchers subsequently used the model with HLA testing to correctly identify patients from the failed clinical trial that did and didn’t develop antibodies to the engineered protein. HLA testing is already common for organ donations, prompting study leader Zuben E. Sauna of FDA to say that more studies like this may lead to HLA tests being applied in “precision medicine,” where drugs can be given to the majority of patients they will help, and withheld from the few that they could harm.