New benchmark shows OpenDDE and Protenix-v2 beat AlphaFold 3 at antibody-antigen docking
josephdviviano · x · 2026-11-16
A standardized benchmark of five antibody-antigen co-folding models (AlphaFold 3, Protenix-v2, OpenDDE, ESMFold2, ESMFold2-fast) finds OpenDDE achieves the highest median interface DockQ and top docking pass rate on 200 complexes, with Protenix-v2 close behind — both statistically outperforming AlphaFold 3 while offering more permissive licenses. The authors also contribute SAbDab2-Bench, 88 post-2024 complexes, for fair comparison across inputs, sampling, ranking and interface metrics.
More from Research
- DeepMind paper: chip sim waveforms as SQLite lets agents ace 95.33% of verification queries — rohanpaul_ai · 2026-11-16
- Anthropic Study: Analysts Follow Their AI Agent's Lean on 86% of Judgment Calls — emmanuelvivier · 2026-11-16
- Axiom Math pitches 'Formalizing All Math' at Caltech Mathathon, one formalization a day — _sathvikr · 2026-11-16
- AI agent swarm of 20 Claude models proves first nontrivial lower bounds for Naslund's crossing problem, formalized in Lean — burny_tech · 2026-11-16
- Dwarkesh Patel: AI math progress matters more for AI research than math — Dwarkesh Patel · 2026-11-16
- Hawkeye brings hardware-aware kernel optimization to coding agents, at NeurIPS — simonguozirui · 2026-11-16