On Thu, Dec 6, 2012 at 4:51 PM, Thomas Evangelidis <tevang3.gmail.com>wrote:
> > > I am trying to use the optimized GBn model (igb=8). Below is the
> sequence
> > > of commands I issue in tleap:
> > >
> > > tleap -s -f
> /home/thomas/Programs/amber12/dat/leap/cmd/leaprc.ff99SBildn
> > > loadAmberParams frcmod.ff99SBnmr
> > > set default pbradii mbondi3
> > > mol = loadpdb x15y120z15.pH6.pqr
> > > saveAmberParm mol structure.prmtop structure.inpcrd
> > > quit
> > >
> > > But when I run energy minimization I get the following error:
> > >
> > > Atom 521 has radius 2.20000000000000 outside of
> > allowed
> > > rang
> > > e
> > > of 1.0 to 2.0 angstroms for igb=7. Regenerate prmtop file with bondi
> > > radii.
> > >
> >
> > What is atom 521? My guess is that it is the CG2 atom of an isoleucine.
> > Previous convention assigned C# atom types (where # is an integer) to
> > nucleic acids, and that assumption appears to have been built into the
> > radii determination for the mbondi3 radius set. Specifically, it sets
> all
> > C3 atom types to radius 2.2, and all other C atom types to 1.7 (which is
> > within the desired range).
> >
> > You can see the radii logic in
> > $AMBERHOME/AmberTools/src/parmed/ParmedTools/changeradii.py. It's
> > significantly more convoluted in the tleap source code.
> >
> > You can fix this in ParmEd with the following commands:
> >
> > changeRadii mbondi3
> > change RADII :ILE.CG2 1.7
> >
> > This will 1) change the radius set to mbondi3, and 2) change all RADII in
> > the mask :ILE.CG2 to 1.7 (which is the 'intended' radius selection for
> > mbondi3).
> >
> > Or, of course, you can use ff12SB or another 'official' Amber force field
> > instead if ff99SBildn.
> >
> > HTH,
> > Jason
> >
> > P.S. ParmEd and tleap should give identical radii for the same keyword,
> > barring of course instances when individual radii were changed by ParmEd.
> >
> >
> Hi Jason.
>
> I guess the indices in AMBER start from 1. I don't want to confuse it with
> VMD where the indexing starts from 0. If the previous is true then atom 521
> is the CG2 carbon of an ILE. Otherwise it is the HG21 hydrogen of an ILE.
> What radius should that hydrogen have? 1.2 A.?
>
I assure you, the atom in question is the CG2 carbon. Amber indexes from
1, VMD indexes (atoms only!) from 0. The ParmEd commands I provided above
should 'properly' set the mbondi3 radii for (all?) ff99SBildn-parametrized
proteins. Specifically, the radii logic defines C1, C2, and C3 atom types
to have a radius of 2.2 (mbondi2 radii -- which are the same as mbondi3
except for 2 or 3 specific atoms). This was supposed to apply to nucleic
acid carbons, I think, but accidentally hits the 'new' ILE atom type
introduced to fix a specific torsion (without breaking others). The
protein carbon atoms were supposed to have a radius of 1.7.
The radii logic can also help provide insight into what each radius was
'intended' to be.
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 Thu Dec 06 2012 - 15:00:02 PST
