posebusters

Multi-stage Riemannian flow matching for physically valid molecular docking, with GNINA scoring, PoseBusters filtering, CLI inference, and benchmarks.

High-throughput docking pose validation: symmetry-corrected RMSD and lightweight PoseBusters-style distance/clash filters.

Unofficial MCP server for PoseBusters – validate molecular poses via HTTP or Spaces using the Model Context Protocol (MCP).

Standalone C++26 PoseBusters-compatible pose validation (NativePoseQC + optional upstream bust bridge). Apache-2.0. Independent of FlexAIDdS.

A measured benchmark of three docking scoring functions on 308 protein-ligand complexes, reporting pose accuracy and physical validity together, with the success rate recomputed after removing the crystal starting conformer and the ligand-centred search box.

ML rescoring of AutoDock Vina poses on the PoseBusters set (RDKit, ProLIF, PyTorch) with a leakage audit: honest negative result