Szybki TK 1.7.2¶
Two new API methods added: SetIntramolecularVdWCutoff and GetIntramolecularVdWCutoff
OESz::OESzybki::GetProteinVdWCutoffhave become deprecated. They have been replaced by
Steepest descent (
OEOptType.SD) method added as an optional optimizer instead of BFGS.
Major bug fixes¶
OESzybkiEnsembleResults now accessible from Java.
Minor bug fixes¶
Reusing the OESzybki object after ligand entropy estimation in the protein environment, for other, non-entropy runs of a ligand inside the same protein, required an extra call to
OESzybki.SetProteinin order to set correctly the new protein-electrostatic model correctly. This is not needed anymore.
In the rare case when ligand is moved out from the protein binding site during entropy estimation the run is terminated with the proper warning. Previously, in such a case the attempts were done to reoptimize the ligand and rebuild the Hessian. Such a strategy was wrong because even if formally successful, it would lead to physically meaningless results. The situation described here can happen when the ligand is poorly docked, or if the X-ray structure used as input is very inaccurate.
When the method
OESz::OESzybki::SetUseCurrentChargeswas used which causes the use of input charges for PB or Sheffield calculations, the molecule which was processed with the
OESzybki.operator()methods, on exit received the mmff94 charges. That behavior has been replaced by retaining the initial charges, and zeroing the molecule’s partial charges when the input molecule had none.
Error message for the case when the input structure generates infinite initial gradients is more informative.