Research map / Probabilistic and scientific modeling / Molecules and proteins
PatchET: Learning Enzyme Temperature Properties Through Patch-Based Neural Architectures
Ziqi Zhang, Runze Yang, Longbing Cao, Zhaohong Deng
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 intra-patch module (local patch attention) (ablation) | PatchET · Temperature optimum | RMSE: 17.139 -> 13.884 | Improved |
| add inter-patch module (conv + attention across patches) (ablation) | PatchET · Temperature optimum | RMSE: 15.508 -> 13.884 | Improved |
| increase intra-patch length from 5 to 25 (kernel 5) (ablation) | PatchET · Temperature optimum | RMSE: 15.278 -> 13.884 | Improved |
| increase patch length from 25 to 50 (kernel 5) (ablation) | PatchET · Temperature optimum | RMSE: 13.884 -> 13.646 RMSE; MAE 9.693 -> 9.817 | Mixed |
| use inter-patch kernel size 5 instead of 3 or 7 (ablation) | PatchET · Temperature optimum | RMSE: kernel 3: 13.991, 5: 13.884, 7: 13.813 at length 25 | Mixed |
| increase attention heads from 1 to 4 (ablation) | PatchET · Temperature optimum | RMSE: 13.922 -> 13.884 | Improved |
| add a fifth attention head (ablation) | PatchET · Temperature optimum | RMSE: 13.884 -> 13.896 | No clear effect |
| use PatchET instead of Seq2Topt | enzyme temperature optimum prediction · Temperature optimum | RMSE: 15.67 -> 13.88 | Improved |
| use PatchET instead of Seq2Topt | enzyme temperature range prediction · Temperature range | mean overlap ratio: 0.683 -> 0.707 | Improved |
| use PatchET instead of Seq2Topt on low-boundary Pearson | enzyme temperature range prediction · Temperature range (low boundary) | Pearson: 0.455 -> 0.473; high boundary Pearson CNN 0.478 vs PatchET 0.514, RNN 0.512 | Mixed |
Closest papers by meaning
- Learning to engineer protein flexibility (ICLR 2025 · similarity 0.85)
- ProteinBench: A Holistic Evaluation of Protein Foundation Models (ICLR 2025 · similarity 0.84)
- Predicting mutational effects on protein binding from folding energy (ICML 2025 · similarity 0.84)
- Generative Enzyme Design Guided by Functionally Important Sites and Small-Molecule Substrates (ICML 2024 · similarity 0.83)
- Sidechain conditioning and modeling for full-atom protein sequence design with FAMPNN (ICML 2025 · similarity 0.83)
- Flexibility-conditioned protein structure design with flow matching (ICML 2025 · similarity 0.83)
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.
