AI splits the job in drug chemistry: XGBoost picks experiments, 3D GNN predicts reactions
bravo_abad · x · 2026-10-10
Minot and colleagues assigned two jobs to different models for predicting where reactions occur on drug-like molecules: an XGBoost ensemble selects which experiments to run by scoring disagreement among members — screening 22,000+ candidates in 0.1 seconds — while geometric graph neural networks, trained on the expanded dataset, use 3D atomic arrangements to predict chemistry. The GNN's edge shows on unfamiliar molecular frameworks where random-split tree models fall short.
More from Research
- EMNLP paper: diffusion LMs internally encode latent denoising timestep representations — gsarti_ · 2026-10-23
- KU Leuven and VIB open two Group Leader professorships in AI for Biology, start 2027 — jpillowtime · 2026-10-23
- Subspace Tuning curbs LLM catastrophic forgetting by freezing principal singular directions — scottniekum · 2026-10-23
- Perceptic launches PercepticOS, an AI workspace for 'scientific superintelligence' — nathanbenaich · 2026-10-23
- EPFL's Bosselut to talk long context as test-time learning at ELLIS seminar — andre_t_martins · 2026-10-23
- DRATS: Adaptive Task Sampling Boosts Worst-Task Performance in Multi-Task RL — Cohere · 2026-10-23