Adding potassium ions to AI structure prediction exposes overfolded disordered regions
- This perspective reviews where AlphaFold-style predictors still struggle: judging confidence, membrane-spanning segments, nucleic acids, alternative conformations, ion and ligand binding, and disordered regions that get folded too neatly.
- Using worked examples, the authors show that including a few explicit potassium ions in AF3 runs on disordered regions or G-quadruplexes can noticeably change the predicted conformation relative to ion-free runs.
- They recommend running sequences both with and without potassium to sample alternatives, reporting confidence scores in papers, and combining tools with more high-quality experimental data, especially for data-poor RNA.
💡 Points Based on the abstract
A shifted conformation is not evidence the ion-bound state is real. AF3 will happily place K+ and reshape the model, so before trusting either run I'd compare against SAXS, NMR or CD data, and check whether the ions sit at geometries typical of G-quadruplex coordination. For a disordered target, I'd treat the with/without-K+ pair as hypotheses for MD triage, not a structure to dock into.
📄 Paper · Journal of Chemical Information and Modeling (ACS) ↗ Deep-Learning Predictions of Biomolecular Structures: Persistent Limitations and New Horizons Extended by Explicit Ion Addition