Re: zCoord-1atom and ** General remark: Subject lines **

From: Jerome Henin (
Date: Tue Apr 10 2007 - 15:28:27 CDT

First, I would like to address a general remark to NAMD-list users:

This has become a high-volume mailing list, so each message naturally gets
less attention. So people should use _very_ explicit subject lines. In this
case, the subject is okay, but still I almost missed it - adding the keyword
ABF would have grabbed my attention. This applies to all types of questions,
you can always find one or two keywords that just work, and less precise,
less useful stuff, like "Help! my namd run is all screwed up".

Now Germán, about your question: there are three things you want to consider
when choosing a set of reference atoms for ABF.

a) Stability
One point of view is that we want the reference has to be "stable", meaning
that it should have enough inertia to define the overall position of the
environment, so for example we might want to include all alpha carbons of the
protein (maybe excluding flexible loops).
Whether you choose e.g. all alpha carbons or all carbonyl carbons or both
whould be almost the same, beacause their centers of mass are virtually the
same and the forces ABF applies on these reference atoms are very small

b) Resolution of local details
There might be local fluctuations of the channel environment with respect to
the bulk of the protein. Using the global reference of point a) above, these
details will be smeared out. If that happens, it might be interesting to use
a small set of residues lining the channel as a reference, to get a
well-resolved profile in that region.

c) Efficiency
More atoms increase the computational overhead due to extra communication
between processors, so it is probably not useful to include many hydrogens or
sidechains if you consider the whole protein.

So you see there is a non-trivial choice to be made, but I hope this helps you
make an informed decision.


On Tuesday 10 April 2007 12:24, German Saavedra wrote:
> Dear NAMD users;
> I am trying to run ABF to get energy profiles of a channel, I am using
> zCoord-1atom to move an ion along the channel.
> I selected as abf1 variable all C types atoms of the protein in a radius
> less than 12 (C type and x^2+y^2<144), and I selected as abf2 variable the
> ion.
> The question is:
> Which is the best atoms to select for abf1 variable?
> I mean, all the atoms?, or only C type? O type?, O and C types? others?
> thanks a lot for your answers,
> Germán

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:44:33 CST