Date: Fri, 18 Jul 2003 16:39:15 -0600 (Mountain Daylight Time)
From: "Thomas E. Cheatham, III"
Subject: Re: AMBER: question about quasih?


> I have two questions about quasih program. The first one is: I run MD
> for a ligand which has 23 atoms. I try to use quasih to get the
> frequencies. If I put nmode less than around 30, it works fine. But when
> it goes to 50 or even greater, it dies. This is the information I get
...
> quasih -natom 23 -f 2 -m 969_mass -x xyz -v vec -nmode 60 -first 1 -last
> 250 < dhna_hp_1a.crd > 1a.out

This is somewhat confusing and took me a little bit to figure out myself.
The value of "nmode" has to be less than or equal to N*3/2. I think the
max nmode you can run in your case is 34 and this is what I would
recommend, i.e. running with all the modes possible.

> Another question is: when I read the manual about quasih, it is said
> that I need to use ptraj to superimpose them first. Then I run ptraj and
> use 'rms' command. I assume this command should superimpose snapshots,
> but I am not sure. Any suggestion?

quashi does not have built in support for superimposing the frames; if you
do not fit them to a common reference frame, then the extra motion (due to
translation/rotation) will be included into the covariance matrix and
effectively increase the entropy.

The best solution is to RMS fit to the average structure; alternatively
you can fit to the first frame. To do the former with ptraj requires two
runs through the trajectory...

Create an average structure:

trajin dhna_hp_1a.crd
center mass origin
rms first mass out rms_to_first_frame.dat
average pdb avg.pdb

Then use this produce a new trajectory RMS fit to the average:been

trajin dhna_hp_1a.crd
trajout dhna_hp_1a_fit.crd
reference avg.pdb
rms reference mass out rms_to_average_frame.dat

This new trajectory (dhna_hp_1a_fit.crd) can be used in the quasih.

Note that *many* frames are necessary to get converged entropy values;
towards this end I would suggest running quasih over various (and
increasing) time windows in your trajectory to look for convergence.
Depending on how flexible the system is, it may take quite some time for
the values to converge.

Good luck,

--tom