Re: [AMBER] single -step transformation TI calculation

From: Victor Ma <>
Date: Fri, 16 Dec 2011 16:43:31 -0600

On Fri, Dec 16, 2011 at 11:44 AM, <> wrote:

> Hi,
> > ifce=1, ifsc=0, crgmask='',scmask=':REF.A1',
> > ifce=1, ifsc=0, crgmask='',scmask=':NEW.B1',
> >
> > or
> >
> > ifce=1, ifsc=0, crgmask='',scmask=':REF',
> > ifce=1, ifsc=0, crgmask='',scmask=':NEW',
> hm, I am suprised this did work at all. scmask has no effect when ifsc=0.
> You are mixing two things here: You switch off the SC potential, but still
> (apparently) use two different structures. The only way this could work in
> my opinion is if REF and NEW have the same number, order and starting
> position of atoms. Since scmask is ignored you are effectively changing
> whatever differences there are in your systems all in one go. These two
> ways are therefore the same calculation, run twice. Otherwise, if your
> systems have different atom numbers, it would probably be bad that this
> ran at all without error message...

That's very true. All atoms in the two structures have the exact same
number, order and starting positions. I tried this approach because I
thought in my case soft-core function is not a must (i have the same number
of atoms in both states). Though I am getting reasonable results out of it
but the rms values seem too large (very likely because of the stability
issue you mentioned below). So I guess for a single-step transformation,
the soft-core potential is probably still preferred even for the same
number of atoms.

Thanks a lot!!


> > So it doesn't seem right to me that I'm getting the same DV/DL values
> when
> > you are just changing atom A1 to atom B1 and when you are changing the
> > whole molecule REF to molecule NEW. Also the rms values are big.
> Neither is suprising, since (I think) you do the same transformation and
> done without SC potentials, it may (very system-dependent!) not be too
> stable. Look at the Free Energy curves to check that.
> > ifce=1, ifsc=1, crgmask='',scmask=':REF.A1',
> > ifce=1, ifsc=1, crgmask='',scmask=':NEW.B1',
> Thats more what I would expect for a setup. Are the rms values any smaller
> now?
> > The final results from the three calculations are close (delG=0.46,
> -1.09,
> > 0.30, expt=-0.51, all in kcal/mol). So among these 3 calculations, is one
> well, each of your result is within 1kcal/mol of the experiment, which is
> not bad at first glance. None of the simulations are better or worse in
> principle, showing that SC potentials are not always needed. The
> differences may well be from sampling, longer runs may make them converge
> closer to each other.
> To better judge the quality of your results, it would be useful to do some
> error analysis, e.g. like in Steinbrecher, Joung and Case, J Comp Chem,
> 2011.
> Kind Regards,
> Dr. Thomas Steinbrecher
> formerly at the
> BioMaps Institute
> Rutgers University
> 610 Taylor Rd.
> Piscataway, NJ 08854
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Fri Dec 16 2011 - 15:00:03 PST
Custom Search