Determine the curation of protonation states in CrossDocked

Determine to what extent protonation states were curated in the original CrossDocked protein–ligand structures, because protonation and charge states depend on experimental and environmental conditions and affect reliable hydrogen reconstruction.

Background

The paper explains that post hoc hydrogen reconstruction for CrossDocked ligands is complicated by inconsistent bond orders, aromaticity assignments, missing charge information, and uncertainty about the environmental conditions represented by the structures. Protonation states influence formal charges, tautomeric forms, hydrogen-bond donor and acceptor assignments, and the placement of explicit hydrogens. Although the structures originate from the Protein Data Bank and are therefore assumed to approximate physiological conditions, the paper notes that this assumption is not guaranteed and leaves the extent of original protonation-state curation unresolved.

References

Although CrossDocked structures originate from the PDB and are therefore assumed to approximate physiological conditions, this assumption is not guaranteed, and it is unclear to what extent protonation states were curated in the original structures.

— NEAT-POCKET: Pocket-Conditioned Autoregressive 3D Molecular Generation with a Neighborhood-Guided Set Transformer  (2609.05097 - Jacob et al., 4 Sep 2026) in Supplementary Information, Section “Post hoc hydrogenation of CrossDocked data”