Research map / Probabilistic and scientific modeling / Molecules and proteins
PROTDYN: A FOUNDATION PROTEIN LANGUAGE MODEL FOR THERMODYNAMICS AND DYNAMICS GENERATION
Yikai Liu, Haoyang Zheng, Lining Mao, Yanbin Wang, Ming Chen, Guang Lin
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 |
|---|---|---|---|
| Use i.i.d. thermodynamics sampling instead of autoregressive dynamics sampling (ablation) | ProTDyn · CATH1 | JSD TICA: 0.206 -> 0.155 (vs Dynamics 100 ns); 0.278 -> 0.155 (vs 10 ns) | Improved |
| Use 100 ns steps with dynamics inpainting instead of 10 ns steps (ablation) | ProTDyn dynamics · CATH1 | JSD TICA: 0.278 -> 0.206 | Improved |
| Use 100 ns steps with inpainting instead of 10 ns steps (ablation) | ProTDyn dynamics · CATH1 MSM | Transition JSD: 0.105 -> 0.055 | Improved |
| Use ProTDyn thermodynamics instead of BioEmu | CATH1 ensembles · CATH1 | JSD RMSD: 0.137 -> 0.012 | Improved |
| ProTDyn trained without octapeptides vs BioEmu trained on them | Octapeptides thermodynamics · Octapeptides | JSD TICA: 0.134 -> 0.207 (BioEmu better on Rg, TICA; tie on RMSD 0.020) | Mixed |
| Smaller time step (10 ns) accumulates autoregressive error (ablation) | ProTDyn dynamics · TICA autocorrelation | autocorrelation decay | Hurt |
Closest papers by meaning
- TEMPO: Temporal Multi-scale Autoregressive Generation of Protein Conformational Ensembles (NeurIPS 2025 · similarity 0.88)
- Ensembits: an alphabet of protein conformational ensembles (NeurIPS 2026 · similarity 0.87)
- Structure Language Models for Protein Conformation Generation (ICLR 2025 · similarity 0.87)
- Aligning Protein Conformation Ensemble Generation with Physical Feedback (ICML 2025 · similarity 0.86)
- ProSST: Protein Language Modeling with Quantized Structure and Disentangled Attention (NeurIPS 2024 · similarity 0.85)
- Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling (ICLR 2024 · similarity 0.85)
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.
