On Tue, Oct 9, 2012 at 6:35 PM, "André C. Stiel" <
andre.stiel.tuebingen.mpg.de> wrote:
> Thanks Jason for the swift reply and obviously for parmed.py...
> I done as told, and it kind of works out... I still have one problem, very
> high vdw and internal energies.
> Let me sum up (for the benefit of others) what I did:
>
> I have a trajectory done with explicit solvent, and now I want to extract
> single residue energies from
> the trajectory. This doesn´t seem to work with the explicit solvent since
> I have to use GB.
>
> So, thanks to Jason, I went this way of preparing my files:
> 1. create a water and ion stripped .mdcrd and .inpcrd with ptraj
> 2. create a equally stripped .prmtop using parmed.py (only in
> AmberTools 12 I figured out ;-)
>
> This works out fine. Now I do the energy calculation using:
>
> #decompose energy on a per-residue basis on the trajectory
> &cntrl
> imin = 5, igb = 5, gbsa = 2,
> ntx = 1, maxcyc = 1,
> ntc = 1, ntf = 3,
> ntb = 0, ntp = 0,
> ntwe = 0, ntpr = 500, ntwx=1,
> cut = 1000.0, idecomp = 1,
> /
> Residues to print
> RES 1 227
> END
> END
>
> What I get is the following. The first frame minimization values look all
> reasonable:
>
> resid |internal |vdw |eel |pol |sas
> ============================================================
> TDC 1 97.818 -4.235 -250.591 72.307 180.515
> TDC 2 71.769 -4.982 -64.649 -3.288 4.889
> TDC 3 88.488 -7.834 -50.514 -91.865 60.060
> TDC 4 91.894 -5.629 -45.914 -97.714 113.038
> TDC 5 61.024 -6.938 -62.792 -4.081 70.127
>
> But in the second frame and following, the internal and vdw jump up to
> unreasonable high values (the original trajectory itself was stable):
>
> resid |internal |vdw |eel |pol |sas
> ============================================================
> TDC 1 1619.028 1018.080 -82.695 9.640 292.780
> TDC 2 1809.027 46.723 -47.603 -2.444 10.724
> TDC 3 57146.677 5799.336 21.396 -87.840 36.338
> TDC 4 26555.148 ********* 254.986 -146.093 96.970
> TDC 5 ********* 44964.405 -18.199 -22.389 83.803
>
> I tried around a little bit with the different igb types (i come to
> understand for 5 i would have to change PBradii), gbsa and the
> cut/rgbmax values.... to no avail, I always get this hight numbers. Might
> it be the problem that my original calculation used ntb=2/ntp=1?
> Or are there other strange things at work here?
>
> Does somebody have suggestions?
>
When you use the "trajout" command in ptraj (or cpptraj) to strip out the
solvent, add the "nobox" command. What is happening, I think, is that the
first 3 coordinates that sander tries to extract from the second frame are
actually the box coordinates of the first frame. After which every atom is
'one' index off, with the first atom having stupid coordinates (i.e., the
box lengths).
This will lead to huge bond energies as well as steric clashes.
HTH,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Oct 09 2012 - 12:30:02 PDT
