Research map / Graph learning / Graph self-supervised learning
An efficient graph generative model for navigating ultra-large combinatorial synthesis libraries
Aryan Pedawi, Pawel Gniewek, Chaoyi Chang, Brandon M Anderson, Henry van den Bedem
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 |
|---|---|---|---|
| sample latents from ex-post MVNormal instead of isotropic prior (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | SA JSD: 0.3340 -> 0.0651 | Improved |
| sample latents from ex-post MVNormal instead of isotropic prior (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | QED JSD: 0.2200 -> 0.0572 | Improved |
| sample latents from ex-post MVNormal instead of isotropic prior (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | MW JSD: 0.4370 -> 0.0970 | Improved |
| sample latents from ex-post MVNormal instead of isotropic prior (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | uniqueness (%): 96.9 -> 99.8 | Improved |
| use mixture of ten normals instead of single normal for ex-post density (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | SA JSD: 0.0651 -> 0.0452 | Improved |
| use mixture of ten normals instead of single normal for ex-post density (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | QED JSD: 0.0572 -> 0.0225 | Improved |
| use mixture of ten normals instead of single normal for ex-post density (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | logP JSD: 0.0643 -> 0.0703 | Hurt |
| use mixture of ten normals instead of single normal for ex-post density (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | MW JSD: 0.0970 -> 0.0686 | Mixed |
| sample latents from ex-post MVNormal instead of isotropic prior (ablation) | CSLVAE · Ex-post density estimation vs uniform library reference | logP JSD: 0.0259 -> 0.0643 | Hurt |
| increase library size at test time versus training subsets (ablation) | CSLVAE · Test-time distribution shift | average likelihood | Hurt |
| CSLVAE vs RationaleRL in-library guarantee | CSLVAE · Enamine REAL | in-library proportion (%): 50.9 -> 100.0 | Improved |
| CSLVAE vs JT-VAE on average likelihood | CSLVAE · Enamine REAL | average likelihood (%): 18.7 -> 72.4 | Improved |
Closest papers by meaning
- Projecting Molecules into Synthesizable Chemical Spaces (ICML 2024 · similarity 0.87)
- LLM-Augmented Chemical Synthesis and Design Decision Programs (ICML 2025 · similarity 0.86)
- CBGBench: Fill in the Blank of Protein-Molecule Complex Binding Graph (ICLR 2025 · similarity 0.86)
- Advancing Retrosynthesis with Retrieval-Augmented Graph Generation (AAAI 2025 · similarity 0.86)
- SynFlowNet: Design of Diverse and Novel Molecules with Synthesis Constraints (ICLR 2025 · similarity 0.86)
- FlashMol: High-Quality Molecule Generation in as Few as Four Steps (NeurIPS 2026 · 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.
