HADDOCK3 2026.10.0

Changes since 2026.8.0.

New features

New [dnascan] module, with code simplification (#1644)
Automatic parameter/topology generation (PRODRG) for ligands embedded in a larger system and for multiple ligands (#1646)
Geometric Watson-Crick and Hoogsteen base-pair detection for all-atom and coarse-grained nucleic acids in restraints generation (#1719)
Workflows are now validated against the general defaults.yaml parameters (#1694)
mpi and grid execution modes added to the general defaults.yaml (#1692)

Per-replica seed offset for sampling_factor in [emref], [mdref] and [flexref] (#1684)
Coarse-graining is now deterministic, with corrected dummy-bead geometry (#1660)
Adenine CG SC1 bead gets its own bead type (#1721)
CG nucleic acid chain break cutoff lowered to 6 Å (#1723)
Better memory size estimate in [contactmap] (#1700)
Cleanup worker pools capped to the available CPUs (#1724)

Default CG nucleobase bead selection for DNA in [caprieval] (#1661)
Glycans: linkage fixes (#1715), alpha-GalNAc (A2G) documentation and NGM C4 stereochemistry (#1710), removal of dead A23a/A26a sialylation loops (#1712)
ssdihed range (#1695)
maxid initialised to 1 (#1696, issue #1392)

Fixes across the notebooks (#1698) and glycan notebook edits (#1667)
Release guidelines added (#1672); AI policy updated with an AI declaration (#1725)

Bumped plotly (7.0.0), kaleido (1.4.0), fastapi (0.141.1), pytest-mock, hypothesis and coverage (#1677, #1705, #1678, #1704, #1706, #1703, #1674, #1675)

Full Changelog: 2026.8.0...2026.10.0