Hello,

I just discovered this this morning. To start the week...

By using the drug abacavir, and I copied/pasted the smiles from the DrugBank website, and PubChem (one gave me a good results on Friday, but I tried the other one this morning and the values were wrong).

So by using molecule from smiles and AlogP, the values are different:

DrugBank NC1=NC(NC2CC2)=C2N=CN(C3CC(CO)C=C3)C2=N1 ;  Alogp = 0.769

PubChem C1CC1NC2=NC(=NC3=C2N=CN3C4CC(C=C4)CO)N ; AlogP = 0.297

Values are not the same!

There is a way to make them similar though:

Molecule from smiles > canonical smiles > molecule from smiles (with the canonical smiles)

Nc1nc(NC2CC2)c3ncn(C4CC(CO)C=C4)c3n1 ; ALogP= 0.769

The problem is that getting the information straight from Pubchem could be considered as good data, with no need of redrawing the molecule.

The differences is due to the aromatic ring, with the double bonds rotated on the ring.

AlogP is done by matching substructure, but I found the results quite confusing.

And it may be useful for other people in the future, in case your AlogP values differ between colleagues with the same compound and the same Pipeline Pilot component.

Cheers,

Jérémy

PS: is it only me or the community website is very very slow today?