Re: [AMBER-Developers] Problems with sleap and iwrap

From: Thomas Cheatham III <tec3.utah.edu>
Date: Fri, 8 Apr 2011 14:46:35 -0600 (Mountain Daylight Time)

> Has anybody experienced problems with sleap and iwrap? (Is anybody using
> sleap routinely?).
>
> Attached is a pdb file and two scripts that 'should' do the same thing in
> sleap and tleap. If you then run MD with the two resulting prmtop and inpcrd
> files things appear to be fine without iwrap=0 but if you set iwrap=1 the
> sleap generated prmtop run behaves VERY strangely.

OK, I just played with this... I do not think it is the prmtop
differences but the initial coordinate differences that are leading to the
behavior, specifically sleap coordinates go from -box/2 to +box/2 (origin)
whereas tleap is going box center (0 to box).

With either prmtop (sleap or tleap) if you use the sleap initial
coordinates it will immediately shift/wrap the protein. [As an aside,
also disconcerting to me was that the coordinates were diverging with the
two prmtops. Then I noticed that the step #0 energies do not match:

NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 7.42 PRESS = -3142.6
 Etot = -26659.6738 EKtot = 151.5632 EPtot = -26811.2370
 BOND = 30.2890 ANGLE = 177.5548 DIHED = 443.3116
 1-4 NB = 202.8250 1-4 EEL = 2154.4513 VDWAALS = 2853.2639
 EELEC = -32672.9326 EHBOND = 0.0000 RESTRAINT = 0.0000
 EKCMT = 47.1277 VIRIAL = 9543.8135 VOLUME = 139959.4531
                                                    Density = 0.7348
 Ewald error estimate: 0.8159E-03
 
 NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 7.42 PRESS = -3142.6
 Etot = -26748.4511 EKtot = 151.5632 EPtot = -26900.0143
 BOND = 30.2890 ANGLE = 88.7774 DIHED = 443.3116
 1-4 NB = 202.8250 1-4 EEL = 2154.4513 VDWAALS = 2853.2639
 EELEC = -32672.9326 EHBOND = 0.0000 RESTRAINT = 0.0000
 EKCMT = 47.1277 VIRIAL = 9543.8135 VOLUME = 139959.4531
                                                    Density = 0.7348
 Ewald error estimate: 0.8159E-03

Angle energies disagree; I have not persued further...]


The problem with glyco imaging (not observed with this test case) must be
that the molecules are not considered "part" of the protein and are
therefore imaged, i.e. in that case the ATOMS_PER_MOLECULE must be
different between tleap/sleap.

--tom


_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Apr 08 2011 - 14:00:08 PDT
Custom Search