Research map / Probabilistic and scientific modeling / Molecules and proteins
GenUnfold: Rapidly Predict Protein Mechanical Unfolding Trajectory via a Physics-Guided Diffusion Model
Yiyuan Zhang, Cailong Hua, VINITENDRA SINGH, Joseph M. Muretta, James M. Ervasti, Murti Salapaka
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ESM-2 coevolutionary context to the standard DiT baseline | DiT-SelfAtten · BSDB test set | Rel l2: 0.4384 -> 0.2165 | Improved |
| add physics-biased attention to the standard DiT baseline | DiT-SelfAtten · BSDB test set | Force-JSD: 0.0832 -> 0.0584 | Improved |
| add co-conditioning (global adaLN plus local cross-attention) with physics-biased attention | DiT-PhyAtten · BSDB test set | FID: 0.0219 -> 0.0117 | Improved |
| use physics-biased attention instead of ESM-2 embeddings for the local stream (ablation) | CrossDiT-ESM-ESM · BSDB test set | Energy-JSD: 0.0533 -> 0.0338 | Improved |
| use physics-biased attention instead of ESM-2 embeddings for the local stream (ablation) | CrossDiT-ESM-ESM · BSDB test set | Rel l2: 0.2017 -> 0.2070 | Hurt |
| add stiffness (K) and resistance (R) graph priors to physics-biased attention (ablation) | GenUnfold · BSDB test set | FID: 0.0178 -> 0.0117 | Improved |
| add stiffness (K) and resistance (R) priors (ablation) | GenUnfold · BSDB test set | Rel l2: 0.2053 -> 0.2070 | Hurt |
| add the resistance map R (ablation) | GenUnfold · BSDB test set | FID: 0.0167 -> 0.0117 | Improved |
| add time-dependent gating (ablation) | GenUnfold · BSDB test set | FID: 0.0128 -> 0.0117 | Improved |
| add CNN-based learning of the physical bias (ablation) | GenUnfold · BSDB test set | Energy-JSD: 0.0378 -> 0.0338 | Improved |
| remove ESM-2 global context (physics only) (ablation) | GenUnfold · BSDB test set | FID: 0.0117 -> 0.0165 | Hurt |
| physics priors plus ESM co-conditioning on unseen CATH superfamilies | GenUnfold vs CrossDiT-ESM-ESM · Unseen-superfamily subset | FID: 0.378 -> 0.088 | Improved |
Closest papers by meaning
- A Physics-Augmented Deep Learning Framework for Classifying Single Molecule Force Spectroscopy Data (ICML 2025 · similarity 0.89)
- Protein Conformation Generation via Force-Guided SE(3) Diffusion Models (ICML 2024 · similarity 0.88)
- Simultaneous Modeling of Protein Conformation and Dynamics via Autoregression (NeurIPS 2025 · similarity 0.88)
- Action-Minimization Meets Generative Modeling: Efficient Transition Path Sampling with the Onsager-Machlup Functional (ICML 2025 · similarity 0.87)
- 4D Diffusion for Dynamic Protein Structure Prediction with Reference and Motion Guidance (AAAI 2025 · similarity 0.87)
- Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling (ICLR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
