Hi, Floris,
Thank you very much for your input. Your result is very valuable for us to
develop better general AMBER force field. As you pointed out, the charges of
N (-1.0) and S (1.4) are too big and the electrostatic interaction is so
strong that the structure is distored when running minimization. When I used
larger dielectric constants or ran GB calculations, I got significantly
improved strcuctures. The following are structure parameters for IGB=5
minimization (500 steps), starting from the mopac minimized structure.
N-S 1.64 1.595
S-H 1.32 1.359
S-O 1.43 1.438, 1.438
N-H 1 1.018, 1.018
N-S-H 101.3 106.76
N-S-O 113.1 101.12, 101,17
H-N-S 108.3 101.65, 101.61
Best
Junmei
On Tue, Jun 17, 2008 at 4:57 AM, Floris Buelens <floris_buelens.yahoo.com>
wrote:
> Dear AMBER users and developers,
>
> I'm interested in free energy calculations on ligand-protein complexes and
> would like to thank the developers for their fantastic work on the AMBER and
> GAFF force fields - I'm in the process of writing up some results for a
> paper which will show very satisfactory agreement with experiment.
> However I have come across an issue with GAFF parameters for sulfonamides,
> for which I'd like to ask for your advice. From my test panel, I observed a
> very significant systematic deviation in binding free energy for a subset of
> sulfonamide-containing compounds. Inspection of MD trajectories shows what
> look like unnatural angles of the H-N-S and N-S-O bonds, both of which are a
> little over 90 degrees, as opposed to the ~110 published for HF/6-31+G*
> calculations (Bharatam et al., Theoretical studies on S ± N interactions in
> sulfonamides, Tetrahedron 58, 2002, 1759 - 1764).
> I've attached a minimised structure of the sulfonamide used in the cited
> study, and here's a quick comparison of the key bond lengths and angles
> (left is Bharatam et al, right is minimised GAFF):
>
> N-S 1.64 1.55
> S-H 1.32 1.36
> S-O 1.43 1.42
> N-H 1 1.01
> N-S-H 101.3 108.47
> N-S-O 113.1 94.38
> H-N-S 108.3 94.39
>
> The most obvious discrepancies seems to be in the N-S-O and H-N-S angles
> (18.7 and 13.9 degrees off), although difference the N-S bond length is also
> significant. This is clearly due to the 1-4 interactions between the
> strongly polarised N-bound hydrogens and S-bound oxygens.
> I've done some initial testing of angles parameters to reproduce more
> closely the H-N-S and N-S-O angles (I set equilibrium angles for both to 120
> degrees and increased the H-N-S force constant by 50%), and this crude
> approach appears to resolve something around half of my observed systematic
> deviation.
> I'd be very grateful if the GAFF developers could take some time to examine
> the sulfonamide moeity more closely, or advise me how to generate parameters
> which do not compromise the force field's consistency?
> Many thanks for your time,
>
> Floris Buelens
> Department of Crystallography, Birkbeck College, London
>
>
> __________________________________________________________
> Sent from Yahoo! Mail.
> A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
to majordomo.scripps.edu
Received on Wed Jun 25 2008 - 06:07:45 PDT
