Pinal: open model that designs proteins from natural-language descriptions of function (Nature; Westlake University)

Pinal (Westlake University, Fajie Yuan lab) is trained on 1.7 billion synthetically annotated protein-text pairs and generates candidate proteins from a plain-language functional description. The Nature paper (published 21 October 2026) reports four wet-lab-validated design tasks (a fluorescent protein, a PET hydrolase, an alcohol dehydrogenase and a metabolic H-protein). Code is Apache 2.0 on GitHub and the pretrained weights are on Hugging Face (westlake-repl/Pinal). No phage angle stated.