From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Fri Jul 19 2013 - 11:50:29 CDT

I would like to add that autoselection of aromatic carbons as both ACC
and DON should be read ACC only (DON, as expected does not converge).

Obvious as the above report is, the very reason of this add is that
numbering of atoms in the list of created files was unexpected. It
does not correspond to the input PDB and resulting psf/pdb. It is not
only different but also difficult to follow. So that one quinone
oxygen missed by the autoselection as ACC is difficult to understand
how to assign it.

I did not use the typical assembling of already parameterized
molecules because of the unusual topology. It also makes difficult to
decide which file to submit.

This should not be read as adverse criticism to FFTK, which actually
is a tremendously useful tool addressing a most difficult task.

francesco pietra

---------- Forwarded message ----------
From: Francesco Pietra <chiendarret_at_gmail.com>
Date: Thu, Jul 18, 2013 at 6:18 PM
Subject: vmd-l: FFTK about Water Interaction
To: vmd-l_at_ks.uiuc.edu

Hi Christopher:
Back to a few months ago:
 From: Mayne, Christopher G (cmayne2_at_illinois.edu)
Date: Wed Mar 27 2013 - 08:24:31 CDT

    Next message: Francesco Pietra: "Re: FFTK Atom types"
    Previous message: Mayne, Christopher G: "Re: FFTK Atom types"
    In reply to: Francesco Pietra: "Fwd: FFTK Atom types"
    Next in thread: Mayne, Christopher G: "Re: FFTK Atom types"
    Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [
attachment ]

On Mar 27, 2013, at 3:34 AM, Francesco Pietra wrote:

> In addition, I wonder whether treating NO as a H-bond acceptor base is
> sound at all. NO lipophilicity is comparable to that of noble gases
> and dinitrogen.

Just because a molecule is lipophilic doesn't meant that it does not
form any favorable interactions with water. I suggest you run the
calculations and see if a minimum interaction energy and distance are
found for either orientation of water.

> In FFTK, autoselection indicates both atoms of NO as
> H-bond acceptors. Even though the Water Interaction module will run, I
> wonder whether chemistry will be taken into account.
>

The algorithm behind the autoselection of H-bond donor and/or acceptor
is admittedly quite simplistic, and was designed to be used as a
starting point for use with more common organics. Users should always
apply their chemical knowledge of their specific molecule to assess
these assignments. That said, for reasonably sized systems, the water
interaction calculation is quite fast. A mis-assigned h-bond
donor/acceptor will result in an abnormally terminated calculation
(does not converge) or unrealistic water position (generally very far
away), so there's not a huge risk associated with mis-assignment--just
exclude these target data from the optimization.

> thanks
>
> francesco pietra

I am now at a ligand of 27 atoms with N-heteroaromatics and
carboxylates. The autofills routine suggests some 17 DON and 24 ACC.
Examining the choice, well it is not unreasonable to suggest
heteroaromatic C-atoms as both ACC and DON. This amounts to 41 QM
calculations, however! Assuming that all these calculations converge,
do you suggest that all that that converges is a MUST? Leaving out
what chemical intuition suggests to be less relevant (e.g., the
interaction of heteroaromatic C-atom with H2O, versus interaction of N
heteroaromatic as acceptor from H2O) will be a bad shortcoming? I
mean, could one focus on the strongest interactions for a molecule
with lipidic and polar parts? Sometimes refining beyond the accuracy
of the tool may turn out as a waste of time.

Thanks

francesco pietra