From: Mayne, Christopher G (
Date: Tue Jul 14 2015 - 09:25:07 CDT


The algorithm behind the autodetection of donor/acceptor is not particularly sophisticated and is only designed to give a quick initial guess. This works out for reasonably simple molecules, but more complete molecules often require manual intervention.

I suspect that the majority of your problem is due to improperly set "element" field in your PDB file, which is used by the algorithm. By default all atoms with element "H" are defined as donors. Non-hydrogens (i.e., "heavy atoms") are subjected to only a slightly more complex decision tree. Heavy atoms with fewer than 4 bonded atoms are default to the acceptors list, while carbons are placed in both lists. The reasoning behind this is that most non-carbon heavy atoms are electronegative (the most common example is oxygen), while the character of carbons with 3 or fewer bonding partners is heavily dependent on the local environment.

Hopefully that sheds some light on what assignments were made, and where you may have errors in your input files that lead to the strange assignments.

As for the water interaction profile with the sulfur atom...a water molecules moving to a large distance indicates that no local minimum was found for the target interaction. In this case it appears that the sulfur atom is acting more as a donor than acceptor, based on your results.

Christopher Mayne

On Jul 14, 2015, at 7:13 AM, fvlieffe wrote:

> hi all,
> background: I'm trying to parametrize the thiamine pyrophosphate by the method divide-and-conquer.
> I divide my molecule in 3 parts: the 4-amino-pyrimidine (6-ring), the thiazolium (5-ring with a positively
> charged N) and the pyrophosphate.
> I already have parameters for the pyrophosphate and I'm working and the 2 former parts.
> Firstly, for the pyrimidine and the thiazolium, I reach the water interaction step in fftk, the step in itself is not difficult to realize
> but after uploading the psf and pdb, come the moment to assign donor or acceptor to the different atoms.
> If I use the autoDetect indices button on fftk, for both (pyrimidine and thiazolium): every atoms are "acceptor".
> It's a bit hard to believe because both contains aliphatic Hydrogen, wich are common donor. So I do not know if I made a mistake somewhere
> or fftk has some issues with that.
> Secondly, I went a little further, generating the gaussian input "donor" AND "acceptor" for all the pyrimidine and the thiazolium.
> After running the gaussian calculation for both molecules (donor and acceptor), the results I obtained for the pyrimidine confirm my believes
> (atoms I expected to be donor and acceptor)
> But in the case of the thiazolium it was more problematic: all the atoms were donors. For example in the thiazolium ring there is a sulfur
> atom, this sulfur atom with the gaussian input "acceptor" has a distance of 31 angtrom with the water molecule when the equilibrium is reached.
> but this sulfur atom with the gaussian input "donor" has a distance of 3.28 angstrom with the water molecule.
> So my question is: when I work with a positively charged molecule (charge on the N of the 5 membered ring) does all the atoms of the ring become
> "donor" for the water interaction step?
> because for my thiazolium it appears to be so.
> thank you
> regards
> --
> Van Liefferinge François
> Phd Student in Chemistry
> S.F.M.B., Université Libre de Bruxelles
> Campus Plaine CP 206/2
> Bâtiment BC, local 1C4.107
> Blvd. du Triomphe
> B-1050 Bruxelles