[gmx-users] May I use "-b" and "- e" in the cosine co n tent calculation of PC?
yanmaocai at 126.com
Mon Sep 11 13:11:04 CEST 2006
Thank you, Tsjerk & David.
Judging from RMSD and other analysis, the system has been in well equilibrium at 5 ns; (however, the cosine contents of last 25 ns and whole 30 ns are also higher than 0.7.) The RMSD during the last 25 nanoseconds (compared to the conformation at 5ns) is within 0.25 nm. After read your letter, I think that the PC1 may be useful, but it cannot represent the global motion of the protein very well because it has not been sampled sufficiently; that is, the significant conformation change observed in PC1 may in fact occur BY ACCIDENT. Did I understand right?
And do you think it will help if I perform multiple MD simulations to test the reproducibility, just as David suggested? (It is quite time-consuming.)
Send: 2006-09-11 18:17:16
From: "Tsjerk Wassenaar" <tsjerkw at gmail.com>
To:"Discussion list for GROMACS users" <gmx-users at gromacs.org>
Subject: Re: [gmx-users] May I use "-b" and "- e" in the cosine con tent calculation of PC?
Hi Mao-Cai Yan,
A high cosine content usually means you're not in equilibrium, or your
trajectory includes the part in which the system is going to
equilibrium. Now, apparently, you're not interested in equilibrium,
but rather in some event of change, which may be required to go from
the starting structure to the equilibrium state or may be a common
transition in equilibrium (in which case you should definitely observe
it more often before claiming anything regarding equilibrium). A high
cosine content in itself is not a qualitative check for your
simulation or fof the principal components extracted from it. It
merely indicates that you're looking at an event of change of your
system, which can be the relaxation from the starting structure or an
undersampled transition, common to your equilibrium state.
To say anything about equilibrium it would also be better to look at
the RMSD from the time averaged structure obtained at the end of the
simulation, which is a much better indicator than the RMSD from the
starting structure. Note that the number of possible conformations
increases rapidly with increasing RMSD, and you'll find the RMSD level
off well before you have true convergence. For larger systems this may
take 15-25 ns or more!
Hope it helps,
On 9/11/06, David van der Spoel <spoel at xray.bmc.uu.se> wrote:
> M. Yan wrote:
> > Thank you very much.
> > The significant conformation changes occurred between 3.3ns and 3.9ns,
> > which I am interested for. The cosine content in this 0.6ns is low (just
> > 0.29; calculated by "g_analyze -f proj5ns.xvg -cc proj5ns_cc.xvg -b 3300<BR>> > -e 3890"); I want to know whether it indicates that the movement of
> > protein IN THIS 0.6ns is believable?
> That depends, since you only see a single event in your 30 ns. It would
> be more convincing if you could show that this is reproducible, by doing
> multiple simulations with different conditions (e.g. velocities,
> temperature, starting structure). If it then happens early on in the
> simulations, and they all converge to the same structure as your 30 ns
> simulation, then it would be convincing...
> David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
> Dept. of Cell and Molecular Biology, Uppsala University.
> Husargatan 3, Box 596, 75124 Uppsala, Sweden
> phone: 46 18 471 4205 fax: 46 18 511 755
> spoel at xray.bmc.uu.se spoel at gromacs.org http://folding.bmc.uu.se
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the gromacs.org_gmx-users