Research map / Probabilistic and scientific modeling / Molecules and proteins
ProMiSE: Protein Multi-State Evaluation Benchmark in Biological Contexts
Bonjae Ku, Seeun Kim, Yubeen Kim, Hahnbeom Park, Chaok Seok
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 |
|---|---|---|---|
| fine-tune only the structure decoding module on MD trajectory data (BioEmu) with frozen pairformer | AlphaFold2-style predictor · ProMiSE intrinsic multi-state | cluster-level success rate: 0.18 (AF3, Boltz-2) -> 0.29 (BioEmu) | Improved |
| use BioEmu decoding instead of AF3/Boltz structure module | BioEmu · ProMiSE intrinsic multi-state | within-distogram std of Struct_holo | Improved |
| condition on apo input (no ligand) vs holo | AF3, Boltz-1, Boltz-2 · ProMiSE ligand-induced | Struct_holo | Hurt |
| higher training-set holo skew (Train_holo) (ablation) | AF3/Boltz structure module · ProMiSE ligand-induced | shift Struct_holo minus DynDisto_holo: positive correlation | Improved |
| stronger MSA holo signal (MSA_holo) (ablation) | AF3/Boltz structure module · ProMiSE ligand-induced | shift Struct_holo minus DynDisto_holo: little to no correlation | No clear effect |
| use model confidence scores to select samples | AF3 and others · ProMiSE intrinsic multi-state | cluster-level success rate | No clear effect |
| use domain-level instead of chain-level training search (ablation) | AF3 vs Boltz-2 · ProMiSE outlier cluster (6crf/7xhq) | Train_holo: chain-level holo bias to Train_holo -0.1148 at domain-level | Mixed |
Closest papers by meaning
- ProteinBench: A Holistic Evaluation of Protein Foundation Models (ICLR 2025 · similarity 0.88)
- Inverse problems with experiment-guided AlphaFold (ICML 2025 · similarity 0.88)
- Unlocking hidden biomolecular conformational landscapes in diffusion models at inference time (NeurIPS 2025 · similarity 0.87)
- CarbonNovo: Joint Design of Protein Structure and Sequence Using a Unified Energy-based Model (ICML 2024 · similarity 0.86)
- Ensembits: an alphabet of protein conformational ensembles (NeurIPS 2026 · similarity 0.86)
- Multiscale guidance of protein structure prediction with heterogeneous cryo-EM data (NeurIPS 2025 · similarity 0.86)
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.
