OpenBind v1

### TL;DR

OpenBind released the first open-source dataset and AI model specifically for structure-based drug discovery. The dataset contains 925 crystallographic binding events from 699 compounds targeting EV-A71 2A protease, with affinity measurements for 601 compounds — providing high-quality protein-ligand data that goes beyond pattern recognition for more reliable molecular predictions.

Key Insights & Metrics

Key Features

• 925 crystallographic binding events from 699 compounds (EV-A71 2A protease focus)
• Affinity measurements (KD values) for 601 compounds via Creoptix WAVEsystem
• Reference benchmarks for docking, cofolding, and affinity prediction methods

Discussion