Re: Simulation of a protein and calcium ions

From: Axel Kohlmeyer (
Date: Wed Oct 09 2013 - 02:45:01 CDT

On Wed, Oct 9, 2013 at 9:39 AM, Stephan Matthias Grein
<> wrote:
> Thank you very much.
> I will investigate these things.
> One problem remains: AutoPSF Buider removes the crystallographically
> resolved Calcium ions from my .psf and .pdb files, why?

like all kinds of "auto" tools it is only meant to deal with simple
tasks. building topologies for biopolymers is dependent on a lot of
heuristics and guesswork. its purpose is to work well in cases, where
no special care needs to be taken and take out some of the tediousness
of the process.

if your case doesn't fall under it, and it very much sounds like it,
you have to do it the old-fashioned way by splitting your system into
pieces, build topology information for each of them with the suitable
information and tweaks and then assembling them back together.

> And how can I
> avoid this?

just use psfgen directly.


> All the best,
> Stephan
> Am 10/8/13 6:37 PM, schrieb Kenno Vanommeslaeghe:
>> I see 3 potential sources of trouble that need to be taken into account:
>> - The simulation time scale, as pointed out by Axel. I would say a 5ns
>> protein simulation is too short for most purposes; something like 50ns
>> would be better.
>> - The force field, as pointed out by Ajasja. In additive force fields,
>> our experience is indeed that without NBFIX, it is impossible to find LJ
>> parameters that both reproduce the RDF in water and the specific binding
>> to protein side chains. It is not impossible that calcium will bind too
>> weakly in the CHARMM force field (though it could be too strongly too).
>> At any rate, getting no binding at all would be a bit surprising.
>> - The ionization state of the side chains! The presence of a Ca2+ will
>> make all basic side chains less basic and acidic side chains more
>> acidic. In nature, weakly acidic side chains can dynamically deprotonate
>> to allow a Ca2+ ion to bind, but in a Class I force field, this is
>> obviously impossible (except when doing constant pH simulations). To see
>> the experimentally observed binding, you may need to try a few
>> ionization states, including ones that would seem slightly unfavorable
>> at neutral pH.
>> On 10/08/2013 04:15 AM, Stephan Matthias Grein wrote:
>> Dear NAMD users,
>> after a couple of weeks learning the NAMD/VMD basics, i came up with
>> one question i could not figure myself:
>> I have generated PSF/PDB files for my protein of interest, using
>> consistent topoplogy and force field parameter files and a consistent
>> NAMD script setup. Solvated this in a waterbox with PBC according to
>> the manual - which works fine.
>> I'm now interested in observing ions moving into binding pockets of
>> the protein (which are reported by literature to be between some EC
>> domains of the protein)... for that i solvated my protein with various
>> concentrations using the AutoSolvation tool in VMD.
>> However, at neither a low nor an extraordinary (unphysiologically)
>> high ion concentration, i could observe ions moving into the binding
>> pockets of the protein and which should stay there, according to
>> literature, at least if the ion concentration is very high.
>> Is there a general failure with my procedure? If yes, would you mind
>> to point it out?
>> Thanks in advance,
>> Stephan

Dr. Axel Kohlmeyer
International Centre for Theoretical Physics, Trieste. Italy.

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:45 CST