[gmx-users] Hydrogen bond autocorrelation

[David van der Spoel]

Den 2019-06-18 kl. 15:58, skrev Gmx QA:
> Dear list,
> 
> This has been discussed many times previously on the list, but I still have
> some questions about hydrogen bond autocorrelation functions.
> 
> I have run a simulation with single molecule of a compound in water which
> can form exactly one inter-molecular hydrogen bond. From the hbnum.xvg
> time-series (i.e. the existence function) of this h-bond I then have a
> python-script to calculate the acf. The script seems to work because the
> resulting acf-function looks exactly the same as the corresponding function
> computed with xmgrace. But it is very different from what I get from gmx
> hbond -ac.
> 
> I read a lot about continuous vs intermittent acfs for h-bonds, the latter
> being called non-continuous in the van der Spoel et al. 2006-paper I think.
> Is the reason for the discrepancy simply that gmx hbond -ac calculates a
> _different_ acf than you get from the hbnum.xvg file (and xmgrace)?
> 
> Then, going forward, after I compute the acf and in the most basic case fit
> to y = a* exp(-t/b), does the value of b correspond to the life-time of the
> h-bond? Or it is the integral (from 0 to inf) that gives you the lifetime?
> This seems to be treated differently by different people.
> 
> Thanks a bunch
> /PK
> 
There are a couple of details indeed, in particular for water. Water may 
bind with oxygen or hydrogens. When your compound provides a hydrogen 
bond acceptor, gmx hbond will compute the hydrogen bond between your 
compound and both H on a water molecule as the same H bond, unless you 
tell it not to with an option -nomerge.

-- 
David van der Spoel, Ph.D., Professor of Biology
Head of Department, Cell & Molecular Biology, Uppsala University.
Box 596, SE-75124 Uppsala, Sweden. Phone: +46184714205.
http://www.icm.uu.se