New preprint: protein foundation models struggle with local chemistry, probed via NMR chemical shifts
anshulkundaje · x · 2026-11-14
A new preprint asks whether protein foundation models truly understand local chemistry, probing them with NMR chemical shifts — per-atom reporters of chemical environment.
Key points:
- The method: NMR shifts provide a quantitative, atom-level readout of how well models capture local electronic and geometric effects
- Findings: prediction errors spike exactly where biology gets interesting — active sites, lid regions, and residues near ligands (a PaDDAH case shows 5 ppm errors around the active triad)
- Implication: current protein models may largely capture sequence/structure statistics rather than real local physical chemistry, a caution for AI-driven protein design
The full details are in the linked preprint.
More from Research
- SFT-ing small models for weight-space programming hits 99.8% on logit lens — lateinteraction · 2026-11-14
- Third-person self-talk regulates emotion without cognitive control, fMRI study shows — AnnaCiaunica · 2026-11-14
- PiHand: latent hand-object interaction priors cut dexterous-hand RL training data 10x — KyleMorgenstein · 2026-11-14
- New paper: symmetry as a common origin of hierarchical emergence and abstraction — mjdramstead · 2026-11-14
- MoE combine traffic cut from 16 to 8.25 bits per value via RHT + S2F6 quantization — snowclipsed · 2026-11-14
- Scientists simulate an entire living cell in 4D, published in Cell — tomchapin · 2026-11-14