3D WHIMS Node - seems to lock up.

I have a list of 700 molecules with SMILES strings. Used CDK-3D Coordinates followed by CDK 3D WHIMS.

CDK-3D Coordinates appears to complete.

CDK 3D WHIMS seems to lock up at 1%, no error messages of any kind.

Any ideas?

Of course, I meant "3D-WHIM" !

I'm already using the nightly build.

I've managed to trace where it gets stuck:

[S-]C#N

(thiocyanate)

Rather peculiar, the next molecule in the list, cyanate [O-]C#N is fine?!

Cheers,

SOH.

Thiocyanate runs fine for me. Just to make sure, have you updated your nightly version recently? If you filter for [S-]C#N and provide only this one row as input for the WHIM node, does it still get stuck? Also, if you copy [S-]C#N into a table using the Table Creator node and connect the 3D coordinates & WHIM node, do you still get the same result?

Thanks,

Stephan

Hi SOH979,

I cannot reproduce your error. Would you mind sharing your list of SMILES or a subset that has the same effect on the node? What CDK version are you on? The nightly might be most suitable for your task.

Many thanks,

Stephan

OK, got the incorrect row; actually stuck on "formate" which was in the data table as SMILES string [H]C([O-])=O

Changing this to C(=O)[O-] solved the problem, though shouldn't the above SMILES still work?

Cheers,

SOH.

Hi SOH,

marvellous, thanks for that. The problem is actually caused by the 3D Coordinates node. It fails to generate valid coordinates for [H]C([O-])=O and inserts n/a values. It works well for C(=O)[O-]  however.

I have patched the problem and updated the nightly. The 3D Coordinates node should create sensible coordinates now.

Hope that helps,

Stephan

OK, thanks for the very quick fix!

Cheers,

SOH.

A few more odd behaviours:

With water SMILES string: O

"3D Coordinates" node fails.

When CORINA Node used to generate 3D coords, "3D WHIM" give very odd output for

several of the properties, e.g. Atomic Polarizabilities.Weta3: ~711 x 10^27

I'm assuming that this is too high !?

Cheers,

Steve.

Hi Steve,

the 3DModelBuilder is not the most robust class in CDK. If you feed "O" to the 3D Coordinates node, you get the followng error message from the force field generator class:

"Could not final configure atom due to problems with force field"

I will file a bug report for CDK.

I get Weta3 0.584 for water -- SDfile from PubChem Compound (with 3D coordinates). Is CORINA proprietary? Would you mind sharing your H2O SDfile?

Cheers,

Stephan

CORINA is proprietary - in fact, I'm using the trial demo at the moment - they have a Knime node.

http://www.molecular-networks.com/products/corina

As far as I could see, the PubChem sdf for water only has 2D coordinates?

The Corina generated 3D coordinate sdf file is attached, rename txt to sdf.

Cheers,

Steve.

Hi Steve,

thanks for sharing the SDfile. I have tried various 3D versions and I always get one 'weird' weta value for water. That appears to be a problem of the class in the core library. I tried other small planar molecules and can reproduce this behaviour.

I will file another bug report for CDK. Hopefully CDK's 3D methods undergo some heavy development at some point.

Sorry not to being able to help more.

Cheers,

Stephan