DivCon Software Release Notes

QuantumBio generally releases its software as a single package referred to as the DivCon Discovery Suite. This Suite of software has gone through many changes over the years as it evolved from its humble beginnings in the academic laboratory to the commercial package that is in place now.

Below is a list of the Release Notes for our MASTER (release version) and DEV (developer version) of the package. Generally, the MASTER version is updated more slowly than our DEV version which is updated on a nightly basis with notable DEV releases. Current licensees have access to both packages along with the nightly build. Visit Software Licensing for more information on licensing the software and Contact Us if you have any questions or to obtain a commercial license.


DEV.1866

Cryo-EM / X-ray QM/MM Refinement:

  • NEW FEATURE: native DivCon QM/MM real-space refinement is now available across the product line and can be used in docking, XModeScore, and structure preparation (protonation). This functionality complements our plugins with PHENIX and BUSTER and can be used independently of these popular platforms or to perform initial preparation. Use –engine DivCon when calling qbphenix or qbbuster or use qbdivcon directly.
  • NEW FEATURE: with the addition of native QM/MM real-space refinement in the DivCon Discovery Suite, cryo-EM is fully supported alongside X-ray. To run XModeScore or QM/MM refinement with cryo-EM *.map files, set –engine DivCon and –experiment cryoEM on the command line. Note: an exception is now thrown in qbphenix/qbbuster/qbdivcon if the incorrect refinement engine is chosen using the –experiment cryoEM command line option.
  • NEW DEFAULT: –makeCIF DivCon (instead of –makeCIF MOE) – to generate all CIFs using QM restraints (note: the final in situ CIF for each ligand / cofactor is automatically generated at the end of each QM/MM refinement).
  • NEW DEFAULT: –protonate DivCon (instead of –protonate MOE) – to protonate the entire structure.
  • Bug Fix: There were some instances in the XModeScore protocol in which resolution was not being properly passed to the –engine DivCon package.
  • NEW FEATURE: final, in situ CIF files are now generated upon completion of the QM/MM refinement. This CIF can be used in subsequent refinements using other refinement engines such as PHENIX or BUSTER.

XModeScore / enumerationDriver:

  • When –exploreChiral is chosen, in addition to sampling chain chiral centers, ligand ring chiral centers are sampled as well.
  • When –exploreFlip is chosen, in addition to sampling planar ring flips, ligand ring-chair (axial ↔ equatorial) flips are sampled as well.
  • Bug Fix: With previous versions, certain tautomers/protomers would not be sampled due to a bug in the protonation of some Nitrogen atoms. This has now been addressed.

Core / Platform Improvements (impacting entire product line):

  • Cleanup Continues (use -v 2 for greater verbosity): removal of extraneous debug output; additional tabular output is provided for interaction energies, structure preparation/protonation, and scoring information; and removal of occasional, spurious error upon completion referring to its failed attempt to close a module.
  • Improved Aromatic Perception: Corrected assignment of certain aromatic GAFF types instead of conjugated types. This modification may impact MTScore results as well as X-ray/Cryo-EM refinement results.
  • Support libraries (e.g. BOOST, EIGEN, HDF5, etc) have been updated on both Linux and Mac to more current versions for greater speed, performance, and stability.
  • The Mac build now uses Apple’s native CLANG compiler instead of GCC. This has led to stability and technical improvements across the package (and for both Linux and Mac) as modifications for CLANG impact GCC as well.
  • The Mac build is properly code-signed as is the installer image. This will allow the “double-click installer” to run correctly on more modern macOS distributions.
  • Bug Fix: in previous versions, there was an incompatibility for certain AMD processors. This was due to the use of an older Intel MKL version which expressed a bug with certain AMD cores. We have addressed this problem and AMD is now supported.

Known Issues:

  • Support for PHENIX 1.21 has not yet been fully tested.
  • When running the DivCon Suite on macOS on an arm64 processor, Rosetta 2 is utilized.

Upcoming Improvements*:

  • Complete threading support for MovableType technologies (scoring, docking, and so on) for even faster free energy analysis support.
  • Multi-blob density docking in XModeScore to sample and score alternative active site locations.
  • Automatic rotamer sampling for missing residue R-groups along with improvements in missing loop preparation for more complete molecular dynamics simulations in MT along with QM/MM-based X-ray/Cryo-EM refinement.
  • We are currently exploring a port of the DivCon Suite to macOS on the arm64 (M2) processor.
  • Master Merge: it has been a while since we merged from DEV to MASTER (for a full release). This has been due to the speed of our development efforts and the numerous additional features associated with engine=DivCon and MovableType. We have several DEV cleanup issues to complete and then we are planning a MASTER merge later in 2024. Until then, it is recommended that you continue to use the DEV package for newer features.

DEV.1785

  • Integrated AM1-BCC charges in MM calculations/optimization/refinement and MTScore (when AMBER pair potential is chosen). In place of default use of PM6 Mulliken charges, charges for unknown/non-standard species follow this path: PM6 optimization, AM1 single point, and finally BCC charge correction. These charges are then averaged according to chemical equivalence and stored as a standard for use in any subsequent MM calculations/optimization/refinement.
  • MOL2 output support has been added for all refinement calculations. If ONIOM (QM/MM) is chosen for refinement, then final charges for the entire QM region(s) is(are) PM6 Mulliken. These charges can be used for subsequent electrostatic analyses.
  • makeCIF=DivCon has now been added as the default CIF-generation tool for any refinement calculations in qbbuster/qbphenix/qbdivcon.
  • protonation=DivCon has now been added as the default protonation tool for any refinement calculations in qbbuster/qbphenix/qbdivcon.
  • engine=DivCon continues to mature and is now used regularly for any refinement and XModeScore calculations in qbbuster/qbphenix/qbdivcon.
  • mmCIF (using GEMMI) support has been added in both input and output. GEMMI is now used for all PDB and mmCIF processing.
  • MTDock (both w/density and w/o density) has more fully matured and can be used for placement of ligands with new protomers/tautomer, flip/rotamer & stereoisomer sampling in enumerationDriver and XModeScore. Use –exploreDocking in qbbuster/qbphenix/qbdivcon.
  • Added support for user-supplied MTZ labels in integrated QM/MM refinement using –engine=DivCon in qbbuster/qbphenix/qbdivcon.
  • Numerous bugfixes and increased maturity throughout product line:
    • Improved QM convergence and removal of geometry distortions in optimization/refinement.
    • DivCon-based protonation using the –prepare qmechanic command line is now fully available and mature for numerous calculations. Refinement will be used for final proton optimization.
    • Mirrored protomers/tautomers, flip/rotamer, and stereoisomer enumeration to alternative copies. 
    • Improved automated molecular perception for edge cases.
    • Improved table information and removal of extraneous screen output.

DEV.1670

New Features / Fixes:

  • Note: MovableType and QM/MM refinement & XModeScore (engine=DivCon, engine=PHENIX & engine=BUSTER) work well for this build. 
  • Numerous bug fixes and further cleanup as the enumerationDriver and density-docker continue to mature.
  • Additional parallelism in enumerationDriver between protomer/tautomer, flip, and stereoisomer sampling and protonation.
  • By default, protomers/tautomers are mirrored to other ligand copies in XModeScore when engine=PHENIX and engine=BUSTER.
  • Client bug fix: there was a bug in cases in which several chains of protein in the input is grouped (e.g. AB) followed by the ligand copies in their own AB chains. So these chains are presented as broken and this would cause errors in the refinement. This build fixes this problem.

Known Problems:

  • We are still working through chirality and flips but this is coming online soon.
  • We are working through some docking odds and ends involving parallelism and proper placement.
  • Finally, there is still significant extraneous debug screen output that is being removed.

Upcoming Improvements:

  • We have added support for multiple-blob docking/scoring and selection and this feature should be switched on in an upcoming package.
  • In MovableType, there is some additional parallelism available in induced fit (etc) MTScoreE and this is now being addressed to further speed up this free energy method.

DEV.1646

New Features / Fixes:

  • QM/MM refinement & XModeScore (engine=DivCon, engine=PHENIX & engine=BUSTER) work well for this build.
  • This build addresses the macOS incompatibility noted in build DivConDiscoverySuite-DEV.1639.

Known Problems:

  • By default, protomers/tautomers are mirrored to other ligand copies in XModeScore. We are still working through chirality and flips but this is coming online soon.

DEV.1639

New Features / Fixes:

  • Note: QM/MM refinement & XModeScore (engine=DivCon, engine=PHENIX & engine=BUSTER) work well for this build and it has been used – on the GridMarkets Cloud – during a recent Workshop at the ACA.
  • –makeCIF DivCon has been added to qbphenix/qbbuster for use of a DivCon-based CIF (instead of ELBOW or GRADE).
  • Upon completion of any refinement process, a final restraint CIF is written associated with the in situ positions/parameters of the ligand.
  • The options –exploreChiral and –exploreFlip have been added to qbphenix/qbbuster to turn on these features in the enumerationDriver respectively. In addition, the user may add atom name or names (within the XModeScore selection) to only explore states around those specific atoms. Otherwise, all states within the XModeScore selection are explored by default.
  • By default, the MOE Score will be calculated and written in the XModeScore table in the event that moebatch is found in the command line PATH. If you don’t see this score (but you have MOE) be sure to add $MOE/bin to your path.

Known Problems:

  • We have found problems with macOS and our new mtz reading/writing facilities. We’re working this through now.
  • By default, protomers/tautomers are mirrored to other ligand copies in XModeScore. We are still working through chirality and flips but this is coming online soon.

DEV.1587

New Features / Fixes:

  • Note: QM/MM refinement & XModeScore (engine=DivCon, engine=PHENIX & engine=BUSTER) work well and this version fixes the bug found on May 11th.
  • This version also has improved table output for ZDD. Finally, opt refinement in enumerationDriver is held to the proper number of atoms in the selection.

Known Problems:

  • We are still also experimenting with –exploreDock to turn on options for placement after enumeration but before the structures go out to BUSTER etc.

DEV.1577

New Features / Fixes:

  • New protonation: the entire protonation algorithm has been updated and improved with better support for new pka prediction and ligand fragments.
  • New QM/MM Realspace refinement (engine=DivCon) for Cryo-EM as well as X-ray crystallography and structure preparation.
  • XModeScore (engine=DivCon) works.
  • And the new enumerationDriver is used for protonation, flip state (–exploreFlip), and chirality (–exploreChiral).
  • New tutorials are available on our GitHub site.

Known Problems:

  • Refinement/XModeScore with BUSTER and PHENIX does not work due to missing gradients in MM region.
  • There is a handful of edge cases that we are going through now which cause some faults and exceptions.
  • We are still also experimenting with –exploreDock to turn on options for placement after enumeration but before the structures go out to BUSTER etc.

DEV.1561

New Features / Fixes:

  • This is the first package to include a more fully functioning internal MTDock routine, MT-based pairwise energy decomposition (MTPWD) for rigid receptor cases, and MovableType with support of torsion optimization before scoring.

Upcoming Improvements:

  • With this MTDock/MTPWD development effort complete and entering the final testing stage, in addition to adding support for MTPWD (with induced-fit or flexible receptor MT), we are merging a new Protonation branch which will yield a more complete internal protonation algorithm and improved support for SMILES along with better support for the new enumerationDriver completed earlier in the year. We will therefore commence merging, testing, and any final development required to deploy engine=DivCon, engine=BUSTER, and engine=PHENIX for QM/MM refinement and XModeScore.

DEV.1526

New Features / Fixes:

  • QM/MM Realspace (engine=DivCon) refinement has been tested and is working well.
  • XModeScore (engine=DivCon & engine=PHENIX) work.
  • The new enumerationDriver is used for protonation, flip state (exploreFlip), & chirality (exploreChiral). New tutorials will be posted to our GitHub soon.

Known Problems:

  • XModeScore (engine=BUSTER) likely works as well but is as yet untested with this package. Update soon. GRADE2 still needs tested support (waiting for the MOGUL license).

Upcoming Improvements:

  • The current focus is completing the in-house protonation tool as it is being updated for correct fragment matching, pKa determination, protonation state sampling, etc.

DEV.1520

New Features / Fixes:

  • QM/MM Realspace refinement has been tested and is working well.
  • This package corrects a final (known) mol2 error.
  • XModeScore (engine=DivCon) works.
  • And the new enumerationDriver is used for protonation & flip state generation.
  • DNA/RNA should be supported as well.

Known Problems:

  • XModeScore (engine=PHENIX, engine=BUSTER) does not work on this package. This is currently being addressed.

Upcoming Improvements:

  • Expanded enumerationDriver for chirality sampling and the in-house protonation tool is being updated to for correct fragment matching, pKa determination, protonation state sampling, etc.

DEV.1504

New Features / Fixes:

  • QM/MM Realspace refinement has been tested and is working well.
  • This package addresses correct mol2 writing (including proper bond type information).
  • It also addresses the previously noted problem with h5 file read/overwrite.
  • XModeScore (engine=DivCon, engine=PHENIX, engine=BUSTER) work.
  • And the new enumerationDriver is used for protonation & flip state generation.

Known Problems:

  • DNA/RNA is not fully supported in typing or mol2 file generation. This is currently being addressed.

Upcoming Improvements:

  • Expanded enumerationDriver for chirality sampling and the in-house protonation tool is being updated to for correct fragment matching, pKa determination, protonation state sampling, etc.

DEV.1484

New Features / Fixes:

  • QM/MM Realspace refinement (with engine=DivCon in qbphenix) has been tested and is working well for this build and there aren’t immediate concerns with this package.
  • DEFAULT CHANGE: we are now able to use a “B” or “C” (etc) atom if it is the only one available.
  • MOL2 files also include proper SYBYL atom types and have been tested with MOE and RDKit.

Known Problems:

  • We are aware of a problem with the DEV package if you have an h5 file with the same name in the directory (it will be reused by default). Until this is addressed the workaround is to DELETE previous H5 files or always run in a clean/empty directory.

Upcoming Improvements:

  • We have also begun addressing issues with engine=PHENIX & engine=BUSTER in qbphenix/qbbuster but use of 3rd party platforms is still not recommended with this package.
  • Also, the new enumerationDriver is almost ready to be deployed in XModeScore.

DEV.1453

New Features / Fixes:

  • QM/MM Realspace refinement has been tested and is working well for this build.
  • We are also addressing some issues with disorder.
  • Finally, the new enumerationDriver (and the ED+MTDock) is coming online and will soon replace the old tautomer/protomer code and support T/P, flip & chiral enumeration + XModeScore.

Known Problems:

  • WARNING: BUSTER & PHENIX do not currently work with the Develop build due to changes in H5 file formatting. Therefore, you should use engine=DivCon for all XModeScore calculations in this build until this compatibility is addressed.

DivConDiscoverySuite-2022-b4882


Note: It is not unusual for Release Notes to include forward looking statements about what is being worked on next and what may be available in a future release. These statements are subject to change based on changing priorities and client feedback.