Re: [Cdk-user] partial charges

[Miguel R. C. <mig...@go...>]

> > The GasteigerPEPEPartialCharges is based on resonance structures. If
> > the method doesn't find any (line 237) and this is the case
> > for c1ccccc1CN what I can see, the method goes out no setting any
> > charge.
> 
> But the ring itself will have resonance structures (?) So shouldn't  
> they be taken into account?

It is true. I put two test
StructureResonanceGeneratorTest.testBenzylamine and
StructureResonanceGeneratorTest.testBenzylamine_Aromatic. And there you
can see that with detection aromaticity the detection of resonance
fails. I added also the test as a bug.

 
> > One of the problems  setting charge into an atom is that  
> > atom.setCharge
> > method does not distinguish between what charge it means: either  
> > sigma,
> > pi or total charge.
> 
> I'm confused as to the semantics of the PEPE class. You note that if  
> I use it, I don't need to use the PEOE class, since it will calculate  
> both. But for my example, it doesn't calculate either of the charges.
> 
Well not exactly. I wanted to say that if you used the PEPE method you
calculate the pi-charge and if you used the PEOE method you calculate
the sigma-charge. The PEPE method internally used the PEOE method to
calculate the sigma partial charge of the resonance structures if are
found one but not the own compound.


> > My suggestion could
> > be to calculate the total charge, have a look to
> > PartialTChargePEOEDescriptor please.
> 
> Aah, thanks. Wouldn't it make sense to have this class as part of the  
> charge package and the descriptor would be a wrapper around it?

For me it is perfect.