[buster-discuss] New Global Phasing BUSTER release (20210716), including first version of Grade2

[Buster Developers]

Dear BUSTER users,

We are pleased to announce a new release of our software, that incorporates a
large number of new features and improvements, together with a new package
(Grade2). As always, every form of feedback is valuable to us: it is very much
appreciated and gratefully acknowledged.

The highlights below are only a subset of the Release Notes, that should be
consulted for full details at

https://www.globalphasing.com/buster/ReleaseNotes/ReleaseNotes-BUSTER_snapshot_20210716.txt


* We are distributing for the first time Grade2, written by Oliver Smart.
  Grade2 is a clean-slate reimplementation of ideas originally used in Grade to
  produce CIF restraint dictionaries by querying the CSD via Mogul, with
  numerous improvements. Grade2 is based around the chemoinformatics toolkit
  RDKit. It uses the Python API to the CSD for Mogul searches and for a custom
  analysis of the search results. The latter enables better use of CSD
  information for rings and planes. Internal testing has shown improvements in
  the description of molecular geometry (via restraint dictionaries) as a result
  of improved estimation of ESD's, of support for more element types, of better
  handling of protonation states and of more sophisticated analysis of planar
  systems and torsion angles. Refinements using Grade2 dictionaries are thus
  expected to behave better and to be more consistent across broad ranges of
  chemistry. Any feedback about the performance of this new tool will be greatly
  appreciated. Grade2 is distributed along with Grade, with both programs
  supported only on operating systems supported by the CSDS.

* A new treatment of hydrogens in refinement - a so-called "hybrid model" - has
  been implemented. It can be activated via the "-M HydrogenHybridModel"
  macro. It will use hydrogens at the nuclear position in the geometry term, and
  at the electron-cloud position in the X-Ray term. The input model could be
  prepared via aB_hydrogenate using the "-full" command-line argument. The
  output file from a BUSTER refinement using this hybrid model will contain
  hydrogens with full occupancy at the nuclear position.

* Fixes have been made to our hydrogenation tool aB_hydrogenate, in functions to
  correct errors from the initial step (Reduce as called by "hydrogenate")
  related to incorrect occupancy values as well as alternate-conformation
  confusion. Thanks to Dirk Reinert for reporting and providing a test case.

* We now use the aB_hydrogenate wrapper tool within buster-report instead of
  directly calling hydrogenate. In this way all the fix-up functionality
  provided by aB_hydrogenate is also available for the PDB file used in the
  ligand analysis via CCDC/Mogul.

* Plots of the log-likelihood and Z-score values on a per-reflection basis at
  the end of a refinement are now included in the X-ray section of
  buster-report. They are generated using the new aB_plot_logLikelihood tool and
  can initially be used for diagnostics. Follow-up work will provide a facility
  for automatically excluding from refinement outliers beyond a given
  log-likelihood threshold.

* An issue on macOS for aB_plot_RecSCC has been fixed, in such a way that plots
  again have meaningful colours. Thanks to Oliver Smart for reporting and
  discussion.

* In order to have individual MolProbity commands (molprobity.clashscore etc)
  behave more similarly to the molprobity.molprobity command (part of CCTBX as
  distributed with CCP4) within buster-report, its input PDB file will first go
  through the new aB_fix_h_occ tool to ensure that zero-occupancy hydrogens are
  changed to full occupancy (i.e. same as that of the parent atom they are
  connected to). This is to prevent those zero-occupancy hydrogens from being
  invisible to e.g. the molprobity.clashscore command while they are modified in
  the same (or very similar) way when passing through the master command
  molprobity.molprobity.

* The keyword setting AddMissingSsbondRecords=yes has been made the default, so
  that missing SSBOND records with unambiguous disulfide geometry (as determined
  via the ana_ssbond tool) will be automatically added. This can be prevented by
  setting AddMissingSsbondRecords=no - or, of course, by providing an input
  model with correct SSBOND information in the first place! NOTE THEREFORE that
  if the lack of correct SSBOND records on input is/was the main reason for
  setting StopOnGellySanityCheckError=no in your "refine" commands or scripts,
  we recommend removing that StopOnGellySanityCheckError flag again (or setting
  it to "yes"). Most checks performed at that stage will then only create a
  stopping error if there is some serious problem with the interpretation of
  connectivity within the input structure, therefore providing potentially
  important feedback to avoid producing distorted structures on output.

* Pipedream will now accept an autoPROC summary.tar.gz file as a valid input for
  the -autoprocdir option.

* The content of the Pipedream JSON output file has been extended. Thanks to
  Raphael Bourgeas and Yorgo El Moubayed for their input into this work.


As always, we kindly ask that support requests and feedback be sent to our
developer mailing lists:

      buster-develop at globalphasing.com    for BUSTER, Pipedream, Rhofit, Grade and Grade2

rather than to individual developers.


With best wishes,

The Global Phasing developers