From:	Don Bashford <bashford "at@at" scripps.edu>
Date:	Wed, 26 Feb 2003 11:50:28 -0800
Subject:	Re: MEAD help

>>>>> "Vlad" == Vlad Cojocaru  writes:

    Vlad> Dear CCLers, I am trying to use MEAD for the first
    Vlad> time. Since the documentation is very scarce I would like to
    Vlad> ask somebody that has already calculated electrostatic
    Vlad> potentials on macromocules using "potential" from MEAD suite
    Vlad> to give me some hints.  I am refering here to a simple basic
    Vlad> calculation. I tried to look at the examples in the
    Vlad> distribution but there is no examples for how to take a
    Vlad> structure from the pdb and calculate the potential and
    Vlad> visualizing it. All the examples deal with more advanced
    Vlad> stuff.  I tried "potential -epsin 1.0 -CoarseFieldOut file
    Vlad> molname" on a pqr file created with ambpdb (AMBER) and a ogm
    Vlad> file that looks like below:
    Vlad>
    Vlad> ON_GEOM_CENT 41 1.0
    Vlad> ON_GEOM_CENT 41 0.25

    Vlad>    I got some warnings and errors that look like this:
    Vlad>
    Vlad> WARNING: AtomChargeSet::get_cuberep: 162charges fall outside of the
lattice,
    Vlad> WARNING: SAVanal_calc: vertex found with count = 2
    Vlad> WARNING: SAVanal_calc: vertex found with count = 2
    Vlad> WARNING: SAVanal_calc: vertex found with count = 1
    Vlad> WARNING: AtomChargeSet::get_cuberep: 2410charges fall outside of the
    Vlad> WARINING from potential main program:
    Vlad> Could not open field point file, comp_test.fpt, for reading.  Exiting
without giving any potentials.

First the easy ones.  The warnings from SAVanal_calc are harmless, and
have to do with numerical degeneracies in the calculation of the
molecular surface.  If you really want to delve into it, see You and
Bashford (1995) JCC 16:743.  The complaint from main about the lack of
an .fpt file is a bug, but also essentially harmless.  Given that you
specified a course field file for output, the program should not
really require you to supply an .fpt file too, and in fact it doesn't
really, the course field file that you requested was generated. (At
least, it is when I try it!)  It's just the warning that's the bug,
and it's on my TO DO list.

The "charges fall outside of the lattice" is potentially more serious.
Taking the coord file you sent me, I see that it's x-coordinates span
about 54 Ang (spans in y and z are less).  The coarse grid that you
specified, 41 points at 1.0 Ang spacing can only span 40 Ang, so some
atoms are bound to fall outside, and that's the source of the first of
the two warnings you get.  In principle, charges outside the box can
be taken care of through the box boundary conditions, but this
requires (in principle) solving already the Poisson eqn for those
charges!  In practice MEAD sets the coarsest-grid boundary conditions
using a simple analytical assumption about the potential on that
boundary due to all charges (inside or outside).  but its validity
depends on that boundary being entirely in the exterior (solvent)
region and relatively far (say, 10 to 15 Ang) from any dielectric
discontinuities.  Bottom line: expand your coarse grid so it spans,
say 80 Ang, either be increasing the number of points or the spacing
between them.

The second "charges fall outside of the lattice" warning is from the
fine grid calculation.  This is harmless if the coarse grid
calculation is good, since the fine-grid boundary conditions are set
up by interpolation from the coarse grid, and therefore include a
reasonable Poisson solution for the influence of the out-of-fine-box
charges on the within-fine-box potential.  However, in your case,
there are two problems: first the coarse-grid calculation is in doubt;
second, because you take the coarse grid as your final program output,
you're not getting any information from the fine grid calculations.
The multiple grids of the .ogm file just represent nesting or
"focusing" as Gilson et al have called it, the fine grid does not feed
information back to the coarse grid as it would in a "multigrid"
calculation.

-Don




Raw Message Text