[Rdkit-announce] 2019.09.1 RDKit Release

[Greg L. <gre...@gm...>]

Dear all,

I'm pleased to announce that the next version of the RDKit - 2019.09 - is
released. The release notes are below.

The release files are on the github release page:
https://github.com/rdkit/rdkit/releases/tag/Release_2019_09_1

Binaries have been uploaded to anaconda.org (https://anaconda.org/rdkit).
The available conda binaries for this release are:
Linux 64bit: python 3.6, 3.7
Mac OS 64bit: python 3.6, 3.7
Windows 64bit: python 3.6, 3.7

Conda builds of the PostgreSQL cartridge are also available:
Linux 64bit: postgresql 9.6, 10, 11
Mac OS 64bit: postgresql 9.6, 10, 11

I believe that conda-forge will also switch to the new version in the near
future.

The online version of the documentation at rdkit.org (
http://rdkit.org/docs/index.html) has been updated.

Some things that will be finished over the next couple of days:
- The conda build scripts will be updated to reflect the new version
- The homebrew script

Thanks to everyone who submitted code, bug reports, and suggestions for
this release!

Please let me know if you find any problems with the release or have
suggestions for the next one, which is scheduled for March/April 2020.

Best Regards,
-greg

# Release_2019.09.1
(Changes relative to Release_2019.03.1)

## Important
- The atomic van der Waals radii used by the RDKit were corrected/updated
in #2154.
  This leads to different results when generating conformations, molecular
volumes,
  and molecular shapes.

## Backwards incompatible changes
- See the note about atomic van der Waals radii above.
- As part of the enhancements to the MolDraw2D class, we changed the type of
  DrawColour from a tuple to be an actual struct. We also added a 4th
element to
  capture alpha values. This should have no affect on Python code (the alpha
  value is optional when providing color tuples), but will require changes
to C++
  and Java/C# code that is using DrawColour.
- When reading Mol blocks, atoms with the symbol "R" are now converted into
  queries that match any atom when doing a substructure search (analogous
to "*"
  in SMARTS). The previous behavior was to only match other dummy atoms
- When loading SDF files using PandasTools.LoadSDF(), we now default to
  producing isomeric smiles in pandas tables.  To reproduce the original
  behavior, use isomericSmiles=False in the call to the function.
- The SMARTS generated by the RDKit no longer contains redundant wildcard
  queries. This means the SMARTS strings generated by this release will
generally
  be different from that in previous releases, although the results
produced by
  the queries should not change.
- The RGroupDecomposition code now removes Hs from output R groups by
default.
  To restore the old behavior create an RGroupDecompositionParameters
object and
  set removeHydrogensPostMatch to false.
- The default values for some of the new fingerprint generators have been
changed so
  that they more closely resemble the original fingerprinting code. In
  particular most fingerprinters no longer do count simulation by default
and
  the RDKit fingerprint now sets two bits per feature by default.
- The SMARTS generated for MCS results using the ringMatchesRingOnly or
  completeRingsOnly options now includes ring-membership queries.

## Highlights:
- The substructure matching code is now about 30% faster. This also
improves the
  speed of reaction matching and the FMCS code. (#2500)
- A minimal JavaScript wrapper has been added as part of the core release.
(#2444)
- It's now possible to get information about why molecule sanitization
failed. (#2587)
- A flexible new molecular hashing scheme has been added. (#2636)

## Acknowledgements:
Patricia Bento, Francois Berenger, Jason Biggs, David Cosgrove, Andrew
Dalke,
Thomas Duigou, Eloy Felix, Guillaume Godin, Lester Hedges, Anne Hersey,
Christoph Hillisch, Christopher Ing, Jan Holst Jensen, Gareth Jones, Eisuke
Kawashima, Brian Kelley, Alan Kerstjens, Karl Leswing, Pat Lorton, John
Mayfield, Mike Mazanetz, Dan Nealschneider, Noel O'Boyle, Stephen Roughley,
Roger Sayle, Ricardo Rodriguez Schmidt, Paula Schmiel, Peter St. John,
Marvin
Steijaert, Matt Swain, Amol Thakkar Paolo Tosco, Yi-Shu Tu, Ricardo
Vianello,
Marc Wittke, '7FeiW', 'c56pony', 'sirbiscuit'


## Bug Fixes:
  - MCS returning partial rings with completeRingsOnly=True
 (github issue #945 from greglandrum)
  - Alternating canonical SMILES for fused ring with N
 (github issue #1028 from greglandrum)
  - Atom index out of range error
 (github issue #1868 from A-Thakkar)
  - Incorrect cis/trans stereo symbol for conjugated ring
 (github issue #2023 from baoilleach)
  - Hardcoded max length of SMARTs string cut of input query for FragCatlog
 (github issue #2163 from 7FeiW)
  - VSA_EState {1, ..., 10} calculated by rdkit doesn't seem correct.
 (github issue #2372 from c56pony)
  - MolStandardize: FragmentRemover should not sanitize fragments
 (github issue #2411 from greglandrum)
  - MolStandardize: combinatorial explosion in Normalizer
 (github issue #2414 from greglandrum)
  - MCS code doesn't return envelope MCS if CompleteRingsOnly is true
 (github issue #2420 from greglandrum)
  - RemoveHs() does not remove all hydrogens.
 (github issue #2422 from paulaju)
  - Incorrect assignment of explicit Hs to Al+3 read from mol block
 (github issue #2423 from greglandrum)
  - Cannot set maxProducts > 1000 in RunReactants
 (github issue #2427 from tduigou)
  - Chem.MolStandardize.standardize.Standardizer drops molecular properties
 (github pull #2431 from lilleswing)
  - Canon::rankMolAtoms results in crossed double bonds in rings
 (github issue #2437 from greglandrum)
  - make boost::iostreams optional
 (github pull #2440 from greglandrum)
  - Fix/rgroup sdf isotope
 (github pull #2449 from bp-kelley)
  - Uncharger incorrectly removing charge from boron anions
 (github issue #2452 from greglandrum)
  - Add java builds to azure devops
 (github pull #2460 from greglandrum)
  - Cart fixes
 (github pull #2462 from jones-gareth)
  - Negative atom map values cause depickling to fail
 (github issue #2465 from greglandrum)
  - Deserialization failures crash Java wrapper
 (github issue #2466 from greglandrum)
  - rdkit.six fix and cleanup
 (github pull #2469 from rvianello)
  - dummy atom queries are flagged as complex
 (github issue #2471 from greglandrum)
  - 3D structure display broken in jupyter notebook
 (github issue #2473 from greglandrum)
  - Inconsistent defaults for nonBondedThresh in MMFF optimization
 (github issue #2475 from greglandrum)
  - Fix/rgroup multiple labels
 (github pull #2481 from bp-kelley)
  - 2D Depiction clipped atom highlighting
 (github issue #2486 from DavidACosgrove)
  - BRICSBuild now passes scrambleReagents to children
 (github pull #2488 from greglandrum)
  - Pattern Fingerprint Issues
 (github issue #2501 from jones-gareth)
  - CMake Error: Wrap directories being used when python build is turned off
 (github issue #2516 from jasondbiggs)
  - - fixes ResonanceMolSupplier bug in perceiving conjugated groups
 (github pull #2517 from ptosco)
  - Fix/mmff threadsafety issues
 (github pull #2518 from bp-kelley)
  - update expected SVG output in cartridge tests
 (github pull #2520 from greglandrum)
  - fix to SDWriter docs
 (github pull #2521 from pstjohn)
  - Fix the azure pipelines builds
 (github pull #2522 from greglandrum)
  - Code cleanups from PVS/Studio
 (github pull #2531 from greglandrum)
  - getAtomNeighbors() and getAtomBonds() not in SWIG wrappers.
 (github issue #2532 from greglandrum)
  - Default sanitizerxn doesn't aromatize if possible
 (github issue #2547 from bp-kelley)
  - Add RDKIT_FILEPARSERS_EXPORT to finishMolProcessing
 (github pull #2551 from d-b-w)
  - Chem.rdFMCS.FindMCS hangs for certain ligand pairs
 (github issue #2581 from lohedges)
  - fix the inclusion path for the targets file (#2584)
 (github pull #2589 from rvianello)
  - Fix inocuous typo/bug in Dative bond matching
 (github pull #2593 from ricrogz)
  - E/Z and CIS/TRANS stereo bonds are incorrectly matched
 (github pull #2596 from ricrogz)
  - Uncharger ignores dications
 (github issue #2602 from greglandrum)
  - Possible Garbage Collection Bug in Pharmacophore Generation
 (github issue #2603 from cing)
  - Uncharger incorrectly neutralizes cations when non-neutralizable anions
are present.
 (github issue #2605 from greglandrum)
  - Bad valence corrections on Pb, Sn
 (github issue #2606 from greglandrum)
  - Pb and Sn should support valence 2
 (github issue #2607 from greglandrum)
  - Uncharger incorrectly modifying a zwitterion
 (github issue #2610 from greglandrum)
  - CanonicalRankAtomsInFragment breakTies doesn't
 (github issue #2611 from bp-kelley)
  - Pattern fingerprint failing substructure condition in very large
molecules
 (github issue #2614 from greglandrum)
  - Memory leak with Chem.Atom() constructor
 (github issue #2639 from AlanKerstjens)
  - Fixes: Atoms in non-standard valences not being properly written to mol
blocks
 (github pull #2646 from greglandrum)
  - C++ MCS code returns a null MCS between methylcyclopentane and
methylcyclohexane
 (github issue #2662 from ptosco)
  - CXSMILES writer has error if mol comes from v3000 molfile
 (github issue #2666 from d-b-w)
  - MolToCXSmiles generates error for empty molecule
 (github issue #2667 from greglandrum)
  - fix a problem with normalize, ringinfo, and fragments
 (github pull #2685 from greglandrum)
  - Error when a squiggle bond is in an aromatic ring
 (github issue #2695 from greglandrum)
  - Cannot combine multiple range queries on a single atom.
 (github issue #2709 from greglandrum)
  - setBondStereoFromDirections() returning incorrect results.
 (github issue #2712 from greglandrum)
  - update supplier documentation to reflect python 3 iterator syntax
 (github pull #2719 from greglandrum)
  - removeHs messing up double bond stereo in partially sanitized molecules
 (github issue #2721 from greglandrum)
  - seg fault in ReactionRunner
 (github issue #2722 from greglandrum)
  - Intermittent test failures for JavaDistanceGeometryTests
 (github issue #2727 from greglandrum)
  - Fixes a bug in TorsionConstraint
 (github pull #2732 from ptosco)
  - Apply fix for #1592 to _MolsToGridSVG
 (github pull #2737 from yishutu)

## New Features and Enhancements:
  - Added rankAtoms to ROMol wrapper and added Java test case
 (github pull #1540 from sroughley)
  - Use van der Waals radii from blue obelisk
 (github pull #2154 from UnixJunkie)
  - add generateDepictionMatching2DStructure() to SWIG wrappers
 (github issue #2239 from greglandrum)
  - Added OptimizeMoleculeConfs with pre-generated force-field
 (github pull #2401 from ptosco)
  - FreeSASA improvements
 (github pull #2402 from ptosco)
  - Speed up symmetrizeSSSR
 (github issue #2403 from d-b-w)
  - Trim whitespace from mol fragment SMARTS and check SMARTS presence
 (github pull #2406 from ricrogz)
  - Run clang-tidy over the entire codebase
 (github pull #2408 from greglandrum)
  - Enable Azure Pipelines builds for CI
 (github pull #2409 from ricrogz)
  - Add RDProps interface to Conformers
 (github issue #2441 from greglandrum)
  - Add minimal JavaScript wrapper
 (github pull #2444 from greglandrum)
  - Fixes annoying warnings on MSVC
 (github pull #2454 from ptosco)
  - add prepareMolsBeforeDrawing option for drawMols
 (github pull #2455 from greglandrum)
  - computeGasteigerCharges quality of life improvement for python api
 (github issue #2480 from bp-kelley)
  - Preserve bond direction in fragmentOnBonds
 (github pull #2484 from greglandrum)
  - SanitizeRxn code and docstring cleanup
 (github pull #2491 from greglandrum)
  - Support XYZ format for output
 (github pull #2498 from e-kwsm)
  - vf2 optimisations
 (github pull #2500 from johnmay)
  - Python wrap enhanced stereo setters
 (github pull #2509 from d-b-w)
  - Fix the azure pipelines builds
 (github pull #2522 from greglandrum)
  - add a script for benchmarking fingerprint screenout and substructure
performance
 (github pull #2523 from greglandrum)
  - make "R" in CTABs an AtomNull query
 (github pull #2528 from greglandrum)
  - Expose SDF loading options to LoadSDF
 (github pull #2534 from bp-kelley)
  - Remove unused ctest: testCanon
 (github pull #2541 from ricrogz)
  - Update maeparser and coordgen versions
 (github pull #2542 from ricrogz)
  - Improved handling of bond stereo in reactions
 (github pull #2553 from ricrogz)
  - Code simplification for fingerprints to np array
 (github pull #2557 from ChrisHill8)
  - Integrate Unique Ring Families from RingDecomposerLib
 (github pull #2558 from greglandrum)
  - Allow providing a bounds matrix to EmbedMol
 (github pull #2560 from greglandrum)
  - Enable SimilarityMaps in C++
 (github pull #2562 from greglandrum)
  - Do not run UnitTestMCS.py::TestTimeout in debug builds
 (github pull #2569 from ricrogz)
  - Expose more drawing methods to Python
 (github issue #2571 from greglandrum)
  - Allow Point2D to be constructed from Point3D
 (github pull #2572 from greglandrum)
  - Allows dative bonds to be drawn
 (github pull #2573 from greglandrum)
  - Allow identification of chemistry problems
 (github pull #2587 from greglandrum)
  - Adds MolFragmentToSmarts to generate smarts for a subset of a Molecule
 (github pull #2594 from d-b-w)
  - Removal of redundant wildcards in SMARTS (Null Atom/Bond Query
combination)
 (github pull #2595 from ricrogz)
  - Support range-based charge queries from SMARTS
 (github issue #2604 from greglandrum)
  - Keep PDB info from Maestro files if available
 (github pull #2619 from lorton)
  - optimization of the MolStandardize code
 (github pull #2621 from greglandrum)
  - Assign stereo bond labels in molecules read from SMARTS
 (github pull #2623 from ricrogz)
  - Automatically load the ipython extensions running in Jupyter
 (github pull #2626 from bp-kelley)
  - draw zero-order bonds
 (github pull #2630 from greglandrum)
  - Updated cartridge documentation
 (github pull #2635 from msteijaert)
  - Add new mol hashing code
 (github pull #2636 from greglandrum)
  - emolecules link updated
 (github pull #2638 from marcwittke)
  - Update maeparser to 1.2.1 and coorgen to 1.3.1
 (github pull #2652 from ricrogz)
  - Get numpy include dir programmatically
 (github pull #2653 from sirbiscuit)
 - Fix long columns pandas
 (github pull #2655 from sirbiscuit)
  - Added AtomComparator.AtomCompareAnyHeavyAtom and test cases to FMCS code
 (github pull #2656 from sroughley)
  - The C++ MCS code generates ambiguous SMARTS strings
 (github issue #2663 from ptosco)
  - add bond-selector info to SVGs
 (github pull #2664 from greglandrum)
  - support writing CXSMILES from the cartridge
 (github issue #2668 from greglandrum)
  - support the new hashing code in the cartridge
 (github pull #2671 from greglandrum)
  - Adds additional capabilities to the minimal JS wrapper
 (github pull #2676 from greglandrum)
  - Add MolHash to Java Wrappers
 (github issue #2677 from sroughley)
  - A bunch of changes to the new fingerprinter code
 (github pull #2679 from greglandrum)
  - Add viewBox to default SVG output
 (github issue #2680 from bp-kelley)
  - Allow Java to see RGroup labels in the std::map wrapper.
 (github pull #2681 from bp-kelley)
  - Update maeparser to v1.2.2.
 (github pull #2682 from ricrogz)
  - Update coordgen to v1.3.2
 (github pull #2686 from ricrogz)
  - Add a drawOptions object to IPythonConsole
 (github pull #2691 from greglandrum)
  - Make StructureGroups editable from Python
 (github pull #2692 from greglandrum)
  - Update documentation
 (github pull #2697 from greglandrum)
  - Make removeHydrogensPostMatch=true the default for RGD
 (github pull #2701 from greglandrum)
  - Eat our own dogfood, Clone is deprecated so use copy
 (github pull #2711 from bp-kelley)
  - The MCS smartsString may still be ambiguous
 (github issue #2714 from ptosco)
  - Add threaded runner for the filter catalog
 (github pull #2718 from bp-kelley)
  - Add Leader picker implementation
 (github pull #2724 from greglandrum)
  - Add consideration of ring fusion to the MCS algorithm
 (github pull #2731 from ptosco)


## Deprecated code (to be removed in a future release):

- The old MolHash code should be considered deprecated. This release
introduces
  a more flexible alternative. Specifically the following pieces will be
removed in a future release:
  - The python functionality
`rdkit.Chem.rdMolHash.GenerateMoleculeHashString()`
  - The C++ functionality directly present in the header file
`GraphMol/MolHash/MolHash.h`