Nature: AI model Pinal designs working proteins from plain-language descriptions
A paper published in Nature on October 21 introduced Pinal, a 16-billion-parameter foundation model that designs candidate proteins from natural-language descriptions of a desired function. It was trained on 1.7 billion synthetically annotated protein-text pairs. The team tested it on four design tasks, including a fluorescent protein and a polyethylene terephthalate (PET) hydrolase, and observed the intended function in each case. One designed H-protein raised product titre 1.7-fold over the native E. coli version in a CO2 fixation pathway.
For biomolecular science students, protein design is moving toward describing goals in words, but experiments still decide what works. Skills in protein biochemistry, enzyme assays and computational tools will be needed to test and refine these designs.
Would you trust an AI-designed protein enough to build a research project around it?

0 Comments
Sign in to join the discussion
No comments yet. Be the first to comment!