NATIVE-ID Plans Large-Scale Protein Aggregation Experiments to Train AI Models
Published in Artificial Intelligence, Biology, Brain/Neurology, Tools.
Publication date: October 9, 2026.
The Stowers Institute announced participation in NATIVE-ID, a research project receiving up to $28.6 million to investigate and predict protein changes associated with neurodegenerative disease. The project is led by the Innovative Genomics Institute at UC Berkeley within ARPA-H’s BIOGAMI program.
Stowers investigator Randal Halfmann’s laboratory will receive approximately $4.1 million over two years to generate experimental data on protein aggregation. Its planned contribution uses DAmFRET, a technique measuring protein self-assembly in individual living cells.
The laboratory aims to test 50,000 proteins expressed separately in yeast under varied conditions, producing more than 10 billion aggregation measurements across over one million samples. These are prospective targets, not completed datasets.
Other collaborators will combine human-neuron experiments, protein-interaction measurements, deep learning, structural studies and computational simulations. The effort addresses intrinsically disordered proteins, whose behavior cannot be described by one stable structure.
The announcement establishes a funded research effort rather than a validated disease-prediction test or treatment. Whether models trained on these experiments can reliably predict harmful changes in human cells will require further evaluation.