Re: MMPBSA-like analysis

From: Hannes Loeffler (Hannes.Loeffler_at_stfc.ac.uk)
Date: Tue Oct 29 2013 - 04:32:01 CDT

On Fri, 25 Oct 2013 16:05:05 -0400
Kenno Vanommeslaeghe <kvanomme_at_rx.umaryland.edu> wrote:

> Good stuff! For the psfgen issue, I think I might have a possible
> solution:
> - CHARMM version 36b1 and newer appear to contain code to read XPLOR
> psf files, so you can read the psfgen psf and thereby sidestep the
> atom ordering issue.
> - If you're stuck with an older CHARMM version, the attached shell
> script should be able to convert an xplor psf into a CHARMM psf that
> can likewise be read into CHARMM (though I haven't thoroughly tested
> it).
>
> One question for Hannes: what was the reason for removing the
> quasiharmonic analysis from the original script at
> http://www.charmm.org/ubbthreads/ubbthreads.php?ubb=showflat&Number=1118
> ?

Personally I am sceptical about the practical applicability of QHA in
the context of how MM-PBSA studies are typically done. In my opinion
it is ok to do comparably short simulation although I think one should
aim at multiple independent simulations to obtain statistically
meaningful results. This assumes of course that no large scale
conformational changes play a role.

But that's exactly what you are doing with QHA, i.e. look at global
motions which are known to converge rather slowly. Only to get all QHA
modes you would need 3N time steps, where N is the number of atoms one
chooses to include in the QHA analysis. So you would need to aim at
time steps >> 3N and ideally do that with multiple runs. In my case it
would have been just possible to do an analysis with the C-alpha
atoms only (I tried and the result was utter rubbish). So I guess
leaving out QHA was also determined by my case.

Hope that helps,
Hannes.

-- 
Scanned by iCritical.

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:53 CST