Re: Full Electrostatic Calculations on Protein in Vacuo

From: Axel Kohlmeyer (
Date: Thu Oct 16 2008 - 10:08:23 CDT

On Thu, 16 Oct 2008, MIke S wrote:

MS> Hello,
MS> I'm simulating a protein in vacuo. The results from the MD simulation will
MS> be compared with my mass spectrometry data. I would like to compute full
MS> electrostatics for the protein so I can accurately simulate the unfolding of
MS> the protein at different charge states. As I understand, full
MS> electrostatics can be implemented using PME, however this requires a

if you want the interactions in vacuo and _no_ interactions
between periodic images, then you must _not_ use PME. a plain
cutoff of half the box size with a sufficiently large box
should do the trick. how accurate that would finally be,
is to some degree a matter of opinion. an interesting question
for examples is: what is the temperature of the molecules/fragments
that you detect in your mass spectrometer? ...and they are passing
through an electric field gradient at high speed, right?

MS> periodic boundary condition. Since I'm working without a solvent and I
MS> don't want to apply an external restraining force is it still possible to
MS> calculate full electrostatic interactions in some other way? If not, I'm
MS> thinking of using a large cutoff value, but I don't know how large I should
MS> make it and whether or not this is even accurate. Wouldn't a large cutoff
MS> value affect the accuracy of the van der Waals calculations (since these are
MS> short-range interactions)? Also, I've noticed that there is a certain

van der Waals interactions decay much faster than coulomb (r^-6
!!) so the error will be small. moreover, unless you are using
a special force field parameterized for simulations in vacuo, you'll
be having a much larger error from the fact that your force field
will include a mean-field type polarization from water molecules
that you don't have in your setup.

MS> limit to the cutoff value that I can use. Beyond this value I get a fatal
MS> error message in my simulation.

for efficiency reasons, NAMD makes certain assumptions about how it
can partition the amount of work and what kind of simulations you
want to do. whenever you try a very unusual setup, you may run into
limitations due to these assumptions.


MS> My plan for now is to use the following parameters to evaluate the
MS> non-bonded interactions:
MS> cutoff 12
MS> switching on
MS> switchdist 10
MS> pairlist 13.5
MS> nonbondedFreq 1
MS> fullElectFrequency 1
MS> With my current knowledge of NAMD this is the logic I'm using: To calculate
MS> van der Waals interactions accurately I will use a short cutoff value with a
MS> switching function to smooth out the van der Waals potential energy. With
MS> the fullElectFrequency parameter full electrostatics will be calculated for
MS> atom pairs beyond the cutoff of 12 angstroms. The configuration file will
MS> not include anything related to PME. So, I'm assuming that by invoking the
MS> fullElectFrequency parameter NAMD will automatically evalute full
MS> electrostatic interactions.
MS> Do you think this would work?
MS> Your suggestions will be more than helpful.
MS> Mike

Axel Kohlmeyer
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
If you make something idiot-proof, the universe creates a better idiot.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:59 CST