Research map / Reinforcement learning / Visual and representation RL
Hit and Lead Discovery with Explorative RL and Fragment-based Molecule Generation
Soojung Yang, Doyeong Hwang, Seul Lee, Seongok Ryu, Sung Ju Hwang
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 |
|---|---|---|---|
| use FREED(PE) instead of REINVENT | docking-guided molecule generation · fa7 | hit ratio: 11.8% (2.6%) -> 26.3% (6.5%) | Improved |
| use FREED(PE) instead of REINVENT | docking-guided molecule generation · fa7 | top 5% score: -9.649 (.186) -> -10.426 (.314) | Improved |
| add pharmacochemical filter reward (ablation) | REINVENT · fa7 | hit ratio: 11.8% (2.6%) -> 8.8% (2.5%) | Hurt |
| add pharmacochemical filter reward (ablation) | REINVENT · fa7 | top 5% score: -9.649 (.186) -> -9.391 (.160) | Hurt |
| add pharmacochemical filter reward (ablation) | MORLD · fa7 | hit ratio: 1.1% (0.3%) -> 1.8% (0.6%) | Improved |
| add pharmacochemical filter reward (ablation) | MORLD · fa7 | top 5% score: -8.353 (.105) -> -8.483 (.148) | Improved |
| use predictive-error prioritized replay instead of vanilla SAC (ablation) | FREED · fa7, parp1, 5ht1b | hit ratio | Improved |
| prioritize replay using predictive error instead of TD error (ablation) | FREED · docking score optimization | hit ratio | Improved |
| use PER rather than curiosity-driven exploration (ablation) | FREED with predictive error or Bayesian uncertainty · docking score optimization | hit ratio | Improved |
| increase fragment library from 91 to 350 fragments (ablation) | FREED(PE) · fa7 | hit ratio: 26.3% (6.5%) -> 9.3% (0.5%) | Hurt |
| increase fragment library from 91 to 350 fragments (ablation) | FREED(PE) · fa7 | uniqueness: 0.723 (0.135) -> 0.914 (0.162) | Improved |
| use fragments from known active compounds instead of random ZINC fragments (ablation) | random fragment generation · fa7 | hit ratio: 7.24% -> 18.48% | Improved |
Closest papers by meaning
- Searching for High-Value Molecules Using Reinforcement Learning and Transformers (ICLR 2024 · similarity 0.89)
- Enhancing Ligand Validity and Affinity in Structure-Based Drug Design with Multi-Reward Optimization (ICML 2025 · similarity 0.88)
- GenMol: A Drug Discovery Generalist with Discrete Diffusion (ICML 2025 · similarity 0.88)
- Controllable 3D Molecular Generation for Structure-Based Drug Design Through Bayesian Flow Networks and Gradient Integration (NeurIPS 2025 · similarity 0.87)
- Uncertainty-Aware Multi-Objective Reinforcement Learning-Guided Diffusion Models for 3D De Novo Molecular Design (NeurIPS 2025 · similarity 0.87)
- MF-LAL: Drug Compound Generation Using Multi-Fidelity Latent Space Active Learning (ICML 2025 · similarity 0.87)
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.
