Vernalis Matched Molecular pairs updates

In the last few weeks (including this evening) we have released a series of updates to our Matched Molecular Pairs nodes, resulting in significant further performance improvements, and some changes to canonicalisation, along with a number of bug fixes.  Currently only in the nightly build - please do give the updates a try and let us know if there are any further problems with them.


Full details from the changelog:

17-Feb-2016 - v1.6.3 (Nightly Only)

  • Moved filtering of fragmentations forwards in process, resulting in significant performance improvements for filtered fragmentations (Unchanging HAC and Changing/Unchanging HAC ratio)
  • Improvements to H-removal algorithm
  • Moved all H-removal handling to fragmentation factory, and removed subsequent duplicate calls to H-removal (further performance improvement)
  • Added check for presence of unassigned double bonds before trying to do any assignment (upto 25% performance improvement)
  • Filter/Splitter nodes no longer pass multicomponent structures
  • Fragmentor nodes no longer fail with multicomponent structures
  • Fixed bug in which certain double bond geometries to near-symmetrical aromatic rings became corrupted causing crashes
  • Fixed bug causing multi-fragment node to only perform two cuts to a bond ([2*]-[1*] value) when the maximum number of cuts was 2
  • Improved garbage collector class to remove possible references to deleted native objexts

12-Feb-2016 - v1.6.2 (Nightly only)

  • Fixed loss of stereointegrity issue in MMP fragmentation
  • Fixed canonicalisation issue resulting from above fix (NB 'Values' will not be differently canonicalised to previously)
  • Fixed graceless failures for some salted forms and apparent successful failures of other salted forms

02-Feb-2016 - v1.6.1 (Nightly only)

  • RCSB PDB Loader nodes make 5 attempts to load each file before failing if unable to load (previously only a single attempt was made, resulting in a missing value in the output)
  • Local PDB Loader now makes 5 attempts to load each, but continues with a WARN to the console and a missing value cell if it fails to load a file
  • Bug fixes to MMP nodes which resulted in loss of double bond geometry (Change in RDKit toolkit behaviour), scrambling of attachment point labels and node failures in certain circumstances
  • Nodes transferred to the KNIME core product (Wait for time, Wait to time, Flow variable Case switch) deprecated


Fantastic that you make these nodes available to the community. But could you possibly upload a workflow that demonstrates the utility of the MMP nodes? It's unclear to me how I can link fragmentation to changes in parameters of the matched pairs (e.g. logP or a pharmacological readout).




Thanks for that feedback - firstly, it is a timely reminder that putting up an example workflow is on my to-do-list, and has been for some time.  I will try to get something together in the next few days, with some particular reference to the points you make regards properties, and will post again when it is there.



Great, am looking forward to it.



I am looking forward that too.


There are now some examples on the example server - see Examples -> 99_Community -> 04_Vernalis