Shuffle the Test Set and Protein AI's Best Scores Collapse
New benchmarking work reports that protein language model scores fall sharply once homologous sequences are kept out of the test set. The problem is well documented.
10 articles filed here
Bonbon, a protein-ligand language model trained on 3.2 trillion tokens, reports new inhibitors for nine drug targets. Here is what the preprint shows, and what to check.
New benchmarking work reports that protein language model scores fall sharply once homologous sequences are kept out of the test set. The problem is well documented.
BiteNetI, a new deep-learning model, out-predicts AlphaFold 3 at finding protein metal-ion binding sites, a blind spot for today's protein language models.
A new bioRxiv preprint, Prot-LAMBDA, teaches protein language models explicit 3D geometry and reportedly matches a model five times its size. Why efficiency, not scale, is the real story.
GenBio AI's AIDO Cell is a 'world model' that simulates a human cell across scales, from DNA to shape. Here's what it does, why it matters, and what still needs proving.
A new Science Advances study uses a protein language model to annotate enzyme 'dark matter' in the gut microbiome, mapping metabolism in health and inflammatory bowel disease.