So, I’ve found the source of the problem - not the stereochemistry, but the ‘Cl’ atom. For reasons I don’t remember, our Mol-block parsing code converts the whole line to uppercase each time, so the line:
M V30 16 Cl 2.18109 3.09803 -0.113906 0
becomes:
M V30 16 CL 2.18109 3.09803 -0.113906 0
The code to look up the atomic weight, which is required for the PMI calculations then fails to find anything for the element with the symbol ‘CL’. I will do a bit more checking to try to figure out if fixing this breaks anything else, add a ‘chloro’ example to the test cases and hopefully get a bug fix pushed out later today or tomorrow.
Thanks again for reporting this, and for your patience and help getting us to the bottom of the problem!
Hi, I’m getting a similar error running a similar workflow - I took the Cl -containing molecules out and still fails, so I removed all the -Br containing ones too and I don’t get the 'NullPointerException anymore, so presumably same issue (and with any other two-letter atom). I do however get another error message - will post on that separately if I can’t get it to run after some playing.
Nick
Yep, I’m afraid so - any 2-letter element symbol will break it at the moment. I’m hoping to get the fix out early next week - I have a few issues to resolve with the forthcoming release of KNIME 4.2 first.
Thanks Steve. I’m following the earlier steps you & Natasja went through and have added the Speedy SMILES Explicit Chirality filter as I have some structures that have the enhanced chiral flags. I’ve connected this to the output of an SDF Reader node (the KNIME Base chemistry one) and it has an output column SMILES containing the SMILES string, but on trying to configure the Speedy SMILES filter I get a ‘Dialogue cannot be opened for the following reason: No column in spec compatible to SmilesValue’? Do I need to use a Vernalis SDF reader?
Nick
You will need to use some sort of conversion node from sdf to SMILES format to do that - either the ‘RDKit from Molecule’ or ‘Canonical SMILES’ nodes from RDKit should be able to do that, or the freely available ‘Molconvertor’ node from the free Marvin/InfoChem nodes. If you go down that route, do make sure you keep the SDF column too though, as otherwise you will loose the 3D coordinates.
Thanks Nick. I have the fix, we just have a strange problem with the community build server at the moment which is preventing us releasing it. The nice people at KNIME.com are working on figuring it out…
There should be no need to upgrade unless you want to. The problem is applying across all our builds at the moment, so that will not make a difference. I will ask the people over at KNIME if they are any nearer understanding the source of the issue.
Sorry @natasja - still stuck with this at the moment. I’ve been away for most of the last 3 weeks, and the guys at KNIME have not managed to track anything down at their end yet. At my end, my work PC is currently refusing to reboot after 3 weeks off. It is high up my priority list to sort this though - I need it working too!
Thanks for your patience. This is now fixed in our nightly build and 4.2 stable builds. The other release (4.1, 4.0, 3.7) will follow hopefully tomorrow. Details at Update to v1.27.0 - New Nodes and Bug Fixes
Thanks for sharing the updates. I’m on 4.2 now and the updated node is working fine. I did notice that I had to convert my SDF file to canonical smiles with RDKIT --> RDKIT conf gen --> PMI. If I just used the SDF file to generate the conformers it fails. Not quite sure why, but perhaps another atom typing issue or something like that. Anyway, glad I can use it again.
Thanks @natasja. If you were able to identify which molecule it is in your incoming sd-file that causes the failure and were able to share it that would be really helpful to see if we can fix that issue for you, but understand if that’s not possible.
Hi there, I am trying to generate the PMI of a collection of 100K members. 3D coordinates generation node works but the PMI seems to fail. I think the workflow is correct but the PMI fails maybe because of the size of the library. Is there a maximum of members? I can read above that there is a limitation on chiral molecules, so what is the max number of chiral molecules tolerated and what is the general number of molecules accepted?
Thank you
Best regards
Iolanda
There shouldnt be any limit to the maximum library size
here is no limit on chiral molecules. The bug above was to do with the handling of Cl and Br atoms - it happened that those molecules were also chiral.
So there is something else wrong - to figure out what I’m going to need some more information:
What version of our plugins do you have installed? (Help -> About KNIME Analytics Platform, then click on ‘Installation Details’ - you should be able to find us somewhere on the Features or Plugins tabs - you need at least v1.27.0 for the above fixes.
What type of input column are you using? (Mol, PDB etc)
What is the error message when the node fails? Is there any additional information in the KNIME console?
Does it run OK for a subset of the molecules in your table? If so, can you use a chunk loop to identify the first molecule that fails (assuming that the answer to 3. above doesnt already tell you the Row ID or input molecule!) If you are able to share the input molecule, please could you do so, as that might also help to pin down the problem. (The answer at Vernalis PMI node failure above - the link is to a specific comment in this thread - can show you how to get to the first broken molecule)
Hi Steve, thank you for the quick reply. I have checked and the plugin version is actually v1.26.0 so perhaps that’s the reason?
The workflow works fine for a small subset of molecules (7 taken from a paper as a ref), so that’s how I started to doubt on the limit and not on other aspects. But ok then I will try to update the plugin and see if the problem gets solved.
Thank you
Best regards
Iolanda
Btw these are the two workflows where you can also see the error on the node:
