[gmx-users] Tryptophan Hbond Bug

[Justin Lemkul]

On 11/15/14 7:24 PM, Alexander Law wrote:
> I may have found a bug in the vmx hbond command. In short the cyclic nitrogen
> within the indole ring on a tryptophan side chain is behaving as both Hbond
> acceptor and donor. The end result of this is a constant Hbond being show in
> the output .xvg. Here are some of the detail which led me to this conclusion,
> it must be said however I am a novice and am most likely doing something
> wrong, regardless there is a problem. note: -nitacc did nothing.
>
> I am looking at a potential hydrogen bond forming between the side chain
> carboxyl group on an aspartate and the cyclic nitrogen in the tryptophan
> indole ring. At first my index groups were the whole residues, but I found
> this produced too many Hbond using the -dan command, I then cut it down to
> the aspartate carboxyl oxygens (2 atoms) and the tryptophan nitrogen and the
> studied hydrogen (2 atoms). This gave 1 continuos Hbond forming across the
> whole 200ns simulation. This is false result, viewing this interaction in VMD
> showed the aspartate fluctuating outside reasonable Hbond distance. These 2
> index groups gave 3 acceptors and 1 donor. 2 of the acceptors are the
> aspartate oxygens, I assumed the donor is the nitrogen but the 3rd acceptor?
> I then used the tryptophan N-H and just the N, this gave 2 acceptors and 1
> donor, using N-H and just the H gave 1 acceptor and 1 donor. Trying to avoid
> the nitrogen altogether I used the asp 2 oxygen atoms and just the hydrogen
> atom from the N-H in tryptophan only gave two acceptors. I am at a loss...
> any advise, questions to investigate more or answers are appreciated!
>

All of this seems consistent with what gmx hbond should be doing.  So revisit 
the original premise and quantify what's going on - your only evidence thus far 
that there should not be a hydrogen bond is via VMD.  What does g_dist tell you 
about the distances between these atoms?  Can you show conclusively that the 
distances are outside what can be considered a hydrogen bond?  Are there any 
other anomalous pairs that reproduce the issue?  There is no reason to think 
that there is a "tryptophan bug," as the calculations are the same, irrespective 
of molecular identity.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================