Yeah, those atom types that you are seeing are the names of the UFF atom
types that the RDKit assigns.
If an atom has a hybridization/charge-state that's not recognized, you'll
get those parameterization errors.
If you're aware of a better way to handle this (or a way to sensibly guess
parameters)
The rdkit sdf files were washed with MOE and then read into rdkitagain
for 3D structure generation.
As there are only a dozen problem cases out of 1.5 million compounds, I just
removed them from my main file and downloaded the mol files from chembl and
double check the structures.
Briii
CHEMBL1796997 is Helium and the same molecule that throws the first UFFTYPER
warnings. Originally I had searched my sdf file only for He1 and found no
hits.
>From the original Goddard paper in 1992, there is only He4+4 atomtype
>described explicitly. I can reason based on the error that RDKIT
Greg,
That fixed it! Thanks so much, that makes a lot more sense now.
-Chris
On Fri, Oct 6, 2017 at 1:25 AM, Greg Landrum wrote:
> Hi Chris,
>
> There's an additional step performed during sanitization that recognizes
> that the implicit H needs to be on the N. The steps of a normal full
> mol
4 matches
Mail list logo