AMBER Archive (2003)

Subject: Re: Comparative MD

From: Thomas Cheatham (cheatham_at_chpc.utah.edu)
Date: Fri Jul 11 2003 - 14:32:35 CDT

> The 2 prmtop files are obviously identical but the 2 prmcrd files present some
> very small differences (i.e. on the seventh digits after the decimal point).
> Using these 2 sets of prmtop and prmcrd files, we ran the same MD-IGB1
> protocole. The 2 MD trajectories are totally different (we follow the backbone
> RMSD between a reference structure and the MD snapshots).
>
> Are you surprised by this ?

I am surprised that LEaP gave two different prmcrd files (with small
differences) but not at all surprised by the divergence in the
trajectories. There is a nice by by Braxenthaler et al. (Moult) that
discusses "chaos" in MD simulation [Proteins (1997) 29,417-425].
Basically, as is well known, small (tiny) differences in the starting
conditions can lead to rapid divergence in the trajectories. In the limit
of infinite sampling, equal ensembles should be sampled, however the
requisite sampling could be much longer than you are willing or can afford
to run (depending on the system size).

So, in summary, there can be differences in trajectories on different
machines, differing numbers of processors, slight diffs in coordinates,
etc that will lead to diverging trajectories...

Good luck with your simulation work,

\ Thomas E. Cheatham, III (Assistant Professor) College of Pharmacy
| Departments of Medicinal Chemistry and of University of Utah
| Pharmaceutics and Pharmaceutical Chemistry 30 South 2000 East, Room 201
| & Center for High Performance Computing Salt Lake City, Utah 84112
|
| e-mail: tec3_at_utah.edu phone: (801) 587-9652 FAX: (801) 585-9119
\ http://www.chpc.utah.edu/~cheatham Offices: BPRP295A / INSCC 418