[buster-discuss] New Global Phasing software release (20260424) providing numerous new features, improvements and fixes
BUSTER Developers
buster-develop at globalphasing.com
Fri Apr 24 14:34:04 CEST 2026
Dear BUSTER users,
We are pleased to announce a new release of our software containing
numerous new features, improvements and fixes across all packages.
A summary of the Release Notes is given below. Considering the time
elapsed since the previous release, this summary has been produced by
stripping off any technical details but without paring down the list
of items, so as to fully represent the extent and diversity of the
work done. You will find the complete and detailed information at
https://www.globalphasing.com/buster/ReleaseNotes/ReleaseNotes-BUSTER_snapshot_20260424.txt
Here is the summary:
#### For BUSTER
* Collaborative work with the Structural Enzymology group of Ashwin
Chari (MPI, Goettingen) on ultra-high-resolution studies of proteins
and nucleic acids (in particular, of trapped reaction intermediates)
has highlighted the need for new high-end capabilities in BUSTER,
that have been implemented and are included in this release.
- The first is an ability to alter the default restraints currently
used in BUSTER, either by modifying or deleting existing ones or
by adding new ones.
This can be invoked by the new '-restraints' argument to both
refine and aB_autorefine, that points to a file containing a set
of instructions defining these restraint modifications - please
see the full Release Notes for a complete specification.
This should allow a very fine-grained control of restraints, where
we envisage users wanting to weaken, strengthen, add or delete as
needed a few specific restraints without the need for potentially
error-prone editing or modification of restraint files. Since this
is a new development we would very much appreciate any feedback
and information about use-cases.
- The second is a new macro, "HighRes.macro", to assist with the
task of setting up very-high-resolution refinements. It adjusts a
number of parameters (including TntBfacMin) that are relevant at
such resolutions, and activates the use of a new implementation of
reciprocal-space summation (instead of FFT methods) for structure
factor and gradient calculations. Run with "-M HighRes" to use it.
- The third is the provision of an initial support for using the
DiSCaMB library for structure factor and gradient computations,
enabling the use of the Transferable Aspherical Atom Model (TAAM)
for a more accurate description of aspherical electron density
around atoms.
The use of the DiSCaMB library can be invoked by a new macro
(activated via the "-M DiSCaMB" argument). In order to use TAAM
the value of the environment variable "BANK_PATH" should be the
full path to a databank that DiSCaMB can use e.g. a Multipolar
Atom Types from Theory and Statistical Clustering (MATTS) databank
obtainable from https://github.com/discamb-project/MATTS . Please
note the licensing conditions in that repository before use.
If no such databank is available, BUSTER will fall back to its own
reciprocal-space summation method using the Independent Atom Model
(IAM). For more details see:
https://www.globalphasing.com/buster/wiki/index.cgi?DiSCaMB
* Besides these high-end capabilities, the following fixes and changes
have been applied:
- The Grade2 path is now the default in aB_covalent_ligand (so the
geometry of the linkage region will be determined with Grade2):
this can be reverted via the
autoBUSTER_CovalentLigand_GradeVersion=1 setting.
- The precision of the DPI parameter can be changed via the
Pdb2Dpi_Precision parameter (default = 3).
- Some additional mmCIF items are put into the PDB file generated by
fetch_PDB_gemmi (for mmCIF-only entries): Rwork and Rfree.
- Support has been added for _chem_comp_bond.value_order as an
alternative to _chem_comp_bond.type in restraints dictionaries.
- Added support for reading EM maps/models into CPDBMAP program (the
unit cell will then be taken from the map file and not the model
PDB file).
- Fixed -noRIG flag behaviour in aB_autorefine (thanks to Ashwin
Chari for reporting).
- Rhofit (as called from within Pipedream) will now be run with the
"-trustplanes" option by default, ensuring that planar groups
specified in the input restraint dictionary are not modified nor
broken up during fitting. The previous behaviour can be restored
using the "-notrustplanes" option.
- The symmetry operator ordering for primitive space groups has been
updated to follow the CCP4 convention (stable across CCP4 versions
6.5 through 9.0 for well over 12 years). This change is invisible
in normal usage but ensures full consistency when comparing
symmetry-related outputs with CCP4 programs.
- Added missing HA atom (and associated geometry) at the N-terminus
of typical amino-acid residues when the N-terminus is protonated.
- Added a missing C+N+H angle restraint for the MXPEPTIDE entry.
- Added non-bonded ADP restraints for both the SIM (similarity) and
RIG (rigidity). Both with very low weight and a distance cutoff of
5 AA.
- Specifying keywords are now correctly interpreted also if they are
comma-separated. Multiple instances of the same keyword appearing
on the same line in the CRD file are also correctly interpreted in
that the last instance takes precedence.
- The Gnuplot binary shipped with BUSTER has been updated with
internal font handling, which avoids potential slow-downs related
to font discovery on some systems.
#### For Grade2
* A new option --ccd has been introduced in Grade2. This option
generates a PDB Chemical Component Dictionary (CCD) CIF file for
the ligand molecule. The --ccd option has been implemented to
assist in using in-house protein-ligand structures for training AI
models, such as Boltz (https://github.com/jwohlwend/boltz).
In tests, Grade2 CCD files work with the Boltz ccd.py processing
script.
For existing refined structures, the utility script grade2_utils
also has a --ccd option. This reads a existing CIF restraint
dictionary and produces a CCD CIF file for the molecule. This
functionality should be compatible with dictionaries created by
other restraint dictionary generators, as well as Grade2.
We would like to thank a member of the Consortium for requesting
this new feature.
* Switched Grade2 to using the Gemmi CIF parser
https://gemmi.readthedocs.io/en/latest/cif.html
for reading and writing CIF files. Thanks to Marcin Wojdyr for
developing Gemmi. Using Gemmi has enabled the cleanup of source
code and facilitates storing more information in the restraint CIF
file.
Because of the change, the restraint CIF file produced by Grade2
has slight formatting differences compared to previous versions
(for example, unknown or null values are output as "?" rather than
"."). Please let us know if this causes you any problems.
* Added a Grade2 documentation chapter "Covalent ligands (PTM)":
https://gphl.gitlab.io/grade2_docs/covalent_ptm.html
The chapter presents an approach to generating restraints for a
ligand covalently bound to an amino acid residue. The approach is
to treat the ligand and amino acid as a single monomer, similar to
that used for post-translationally modified amino acids (PTMs).
* Improved the error message written if Grade2 is used with an
incompatible version of the CSD to include a suggestion to check
the online Grade2 / CSD compatibility page:
https://gphl.gitlab.io/grade2_docs/csd_compatibility.html
* Grade2 now works with the CSD release 2025.2 and following,
including the recently released CSD 2026.1. CSD release 2025.2
unexpectedly altered the Python version of the included Miniconda
environment to 3.11, introducing an incompatibility with the
Conda-packed environment used for Grade2. This issue has been
resolved, and current Grade2 release 1.8.0 is compatible with all
CSD releases from 2024.1 to 2026.1.
We will endeavour to ensure that a similar incompatibility does not
arise again.
Before installing a new version of the CSD, please check:
https://gphl.gitlab.io/grade2_docs/csd_compatibility.html
to make sure that the new version works with Grade2 and other
Global Phasing Tools.
The CSD update procedure is irreversible and downloads the latest
version. Hence, if you have sufficient disk space, it is advisable
to install new releases of CSD-core separately rather than using
the CSD update procedure. If there are compatibility problems with
the latest CSD release, then please contact the CCDC, who may be
able to supply an (offline) installer for an older CSD release.
* Grade2 can read recent AceDRG restraint dictionaries that have
altered chem_comp_bond data items: _chem_comp_bond.value_order
rather than _chem_comp_bond.type and
_chem_comp_bond.pdbx_aromatic_flag rather than
_chem_comp_bond.aromatic.
The fix also means that the buster-report tool, Mogul analysis for
ligands, will be able to use the restraint dictionary information.
Thanks to a Consortium user for reporting this bug.
* We fixed a bug where the chem_comp_atom.type_energy for nitrogen
atoms participating in two aromatic rings, such as in PDB component
OF8 (https://www.rcsb.org/ligand/OF8), was incorrectly set to N
instead of NR56, NR66 or NR55.
Thanks to a Consortium user for reporting this bug.
* Fixed a bug where grade2_tests reported a fail in
test_pdb_ccd_nml_using_mogul when CSD 2025.1 was used. The test
has been updated to verify that the PDB component DZ3
(https://www.rcsb.org/ligand/DZ3) has a plane restraint applied to
the amide group, as this is more relevant to pharmaceutical
compounds.
* Fixed a bug where the default Grade2 --lookup feature stopped
working because of changes at PubChem and to PubChemPy
(https://github.com/mcs07/PubChemPy/).
-----------------------------------------------------------------------
The complete and detailed information can be found in the Release Notes
referred to at the beginning of this announcement, that you are invited
to consult in full.
-----------------------------------------------------------------------
As always, we kindly ask that all support requests and feedback
be sent to our developers' mailing list:
buster-develop at globalphasing.com
rather than to individual developers.
With best wishes,
The Global Phasing developers
More information about the buster-discuss
mailing list