Research map / Image generation / Autoregressive and latent generation
Shrinking Proteins with Diffusion
Ethan Baron, Alan Nawzad Amin, Ruben Weitzman, Simon d'Oelsnitz, Debora Susan Marks, Andrew Gordon Wilson
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 |
|---|---|---|---|
| scale SCISOR from small to large | SCISOR · ProteinGym single deletions | Spearman correlation: 0.332 -> 0.573 | Improved |
| scale SCISOR from small to medium | SCISOR · ProteinGym multiple deletions | Spearman correlation: 0.268 -> 0.520 | Improved |
| scale SCISOR from medium to large | SCISOR · ProteinGym multiple deletions | Spearman correlation: 0.520 -> 0.492 | Hurt |
| use SCISOR L instead of PoET with family information | PoET (200M) · ProteinGym single deletions | Spearman correlation: 0.551 -> 0.573 | Improved |
| use SCISOR M instead of PoET | PoET (200M) · ProteinGym multiple deletions | Spearman correlation: 0.488 -> 0.520 | Improved |
| use SCISOR L instead of PoET on activity assays | PoET (200M) · ProteinGym single deletions (activity assays) | Spearman correlation: 0.664 -> 0.604 | Hurt |
| shrink with SCISOR instead of Raygun at 20% deletion | Raygun · RalA shrinking, 20% deleted | designs with TM above 0.5 (out of 5): 1 -> 5 | Improved |
| shrink with SCISOR instead of ProGen2 at 20% deletion | ProGen2 · RalA shrinking, 20% deleted | designs with TM above 0.5 (out of 5): 0 -> 5 | Improved |
| shrink with SCISOR instead of ProGen2 at 20% deletion | ProGen2 · RalA-Sec5 binding, 20% deleted | AlphaFold ipTM: 0.36 -> 0.58 | Improved |
| shrink with SCISOR instead of ProGen2 by 50% on functional site enrichment | ProGen2 · UniProt shrinking, 50% deleted | functional site enrichment | Hurt |
| replace independent-effect ProGen2 baseline with costlier O(L*M) ProGen2 (ablation) | ProGen2 · UniProt shrinking | pLDDT, TM, RMSD, functional enrichment | No clear effect |
| use SCISOR deterministic greedy shrinking vs ProGen2 | ProGen2 · UniProt shrinking (greedy) | pLDDT, TM, RMSD, functional enrichment | Mixed |
Closest papers by meaning
- DePLM: Denoising Protein Language Models for Property Optimization (NeurIPS 2024 · similarity 0.87)
- Diffusion on Language Model Encodings for Protein Sequence Generation (ICML 2025 · similarity 0.86)
- Fine-Tuning Discrete Diffusion Models via Reward Optimization with Applications to DNA and Protein Design (ICLR 2025 · similarity 0.86)
- Steering Masked Discrete Diffusion Models via Discrete Denoising Posterior Prediction (ICLR 2025 · similarity 0.85)
- Diffusion Language Models Are Versatile Protein Learners (ICML 2024 · similarity 0.85)
- Steering Generative Models with Experimental Data for Protein Fitness Optimization (NeurIPS 2025 · 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.
