Welcome to PoseBench’s documentation!¶
Comprehensive benchmarking of protein-ligand structure prediction methods
⚠️ Notice: We have discovered a bug in version
1.0.0affecting the ligand scoring results of each primary-ligand dataset (Astex Diverse, DockGen-E, and PoseBusters Benchmark). We have addressed this bug in version1.1.0, which reports (~15% on average) reduced performance for each method currently in the benchmark. Please rerun your analyses if you have developed code on top ofPoseBench. Thank you for your understanding!
Contents:
- Installation
- Tutorials
- How to prepare PoseBench data
- Downloading Astex, PoseBusters, DockGen, and CASP15 data
- Downloading benchmark method predictions
- Downloading benchmark method interactions
- Downloading sequence databases (required only for RoseTTAFold-All-Atom inference)
- Downloading PDB metadata
- Predicting apo protein structures using ESMFold (optional, preprocessed data available)
- Available inference methods
- How to run a sweep of benchmarking experiments
- How to run inference with individual methods
- How to run inference with
DiffDock - How to run inference with
FABind - How to run inference with
DynamicBind - How to run inference with
NeuralPLexer - How to run inference with
FlowDock - How to run inference with
RoseTTAFold-All-Atom - How to run inference with
Chai-1 - How to run inference with
Boltz - How to run inference with
AlphaFold 3 - How to run inference with
AutoDock Vina - How to run inference with
TULIP
- How to run inference with
- How to run inference with a method ensemble
- How to create comparative plots of inference results
- For developers
- Acknowledgements
- Citing this work
- Bonus