.. _section_szybki_releasenotes_172:

Szybki TK 1.7.2
================================================================================

New features
--------------------------------------------------------------------------------
* Two new API methods added: `SetIntramolecularVdWCutoff` and
  `GetIntramolecularVdWCutoff`

* Methods ``OESz::OESzybki::SetProteinVdWCutoff`` and 
  ``OESz::OESzybki::GetProteinVdWCutoff`` have become deprecated. They have
  been replaced by ``OESz::OESzybki::SetIntermolecularVdWCutoff`` and 
  ``OESz::OESzybki::GetIntermolecularVdWCutoff`` respectively.

* Steepest descent (:oe:const:`OESz::OEOptType::SD`) method added as an
  optional optimizer instead of BFGS.


Major bug fixes
--------------------------------------------------------------------------------

.. oe:language:: python

   * :ref:`OESzybkiEnsembleResults <OESz_OESzybkiEnsembleResults>` now accessible from Python.

.. oe:language:: java

   * :ref:`OESzybkiEnsembleResults <OESz_OESzybkiEnsembleResults>` now accessible from Java.

.. oe:language:: csharp

   * :ref:`OESzybkiEnsembleResults <OESz_OESzybkiEnsembleResults>` now accessible from C#.


Minor bug fixes
--------------------------------------------------------------------------------

* Reusing the :ref:`OESzybki <OESz_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 :oe:meth:`OESz::OESzybki::SetProtein` in 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::SetUseCurrentCharges`` was used which causes
  the use of input charges for PB or Sheffield calculations, the molecule which was processed
  with the :oe:meth:`OESz::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  That behavior has been replaced by retaining the initial charges, and zeroing the molecule's 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. 

