Research map / Image generation / Fast diffusion sampling
MacroGuide: Topological Guidance for Macrocycle Generation
Alicja Maksymiuk, Alexandre Duplessis, Michael M. Bronstein, Alexander Tong, Fernanda Duarte, Ismail Ilkan Ceylan
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 MacroGuide topological guidance to pretrained MolDiff | MolDiff · GEOM-Drug unconditional | macrocycle fraction: 0.053 -> 0.997 | Improved |
| add MacroGuide to MolSnapper | MolSnapper · protein-conditioned | macrocycle fraction: 0.003 -> 0.995 | Improved |
| finetune MolDiff on macrocycles instead of guidance | MolDiff · GEOM-Drug unconditional | macrocycle fraction: 0.851 vs 0.997 for MacroGuide | Mixed |
| finetuned model under protein conditioning | MolSnapper (finetuned) · protein-conditioned | macrocycle fraction: 0.851 -> 0.180 | Hurt |
| naive ring-forcing guidance on a fixed atom subset | MolDiff · GEOM-Drug unconditional | macrocycle fraction: 0.053 -> 0.370 | Mixed |
| initialize denoising with torus-shaped noise | MolDiff · GEOM-Drug unconditional | macrocycle fraction: 0.053 -> 0.054 | No clear effect |
| add MacroGuide | MolDiff · GEOM-Drug unconditional | PoseBusters all tests: 0.663 -> 0.805 | Improved |
| add MacroGuide | MolSnapper · protein-conditioned | PoseBusters all tests: 0.440 -> 0.575 | Improved |
| add MacroGuide | MolSnapper · protein-conditioned | pharmacophore satisfaction: 0.769 -> 0.789 | Improved |
| add H0 death guidance at n=110 atoms (ablation) | MolDiff · large-molecule generation | connectivity: 20% -> 80% | Improved |
| apply guidance every k steps (k up to 5) (ablation) | MacroGuide · protein-conditioned | macrocycle rate and quality: cost divided by at least 5, less than 15% degradation | Mixed |
| start guidance only in late denoising stages (ablation) | MacroGuide · macrocycle generation | macrocycle rate | Hurt |
Closest papers by meaning
- Chain-of-Generation: Progressive Latent Diffusion for Text-Guided Molecular Design (NeurIPS 2026 · similarity 0.87)
- DMol: A Highly Efficient and Chemical Motif-Preserving Molecule Generation Platform (NeurIPS 2025 · similarity 0.87)
- DiSCO: Diffusion Schrödinger Bridge for Molecular Conformer Optimization (AAAI 2024 · similarity 0.86)
- BindGPT: A Scalable Framework for 3D Molecular Design via Language Modeling and Reinforcement Learning (AAAI 2025 · similarity 0.86)
- CBGBench: Fill in the Blank of Protein-Molecule Complex Binding Graph (ICLR 2025 · similarity 0.86)
- Projecting Molecules into Synthesizable Chemical Spaces (ICML 2024 · 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.
