AMBER Archive (2003)

Subject: Re: AMBER: Compiling AMBER, ie, number of atoms limit

From: Robert Duke (rduke_at_email.unc.edu)
Date: Thu Oct 09 2003 - 17:18:27 CDT

Guys - Big solutes get you into trouble in sander 6 / sander 7 using the
sander 6 prmtop. The problem has to do with the representation of the atoms
in dihedrals, where a negative sign can get thrown into the mix to represent
an improper torsion. Basically, atoms above 33,332 had better not be found
in an improper torsion, or grief may follow. However, most the time sander
7 gets away with using the old style prmtop, even with atom counts in the
range of 100,000-200,000, because the bulk of the high atoms in such systems
are solvent (typically water) and possibly ions, both of which don't have
dihedrals. In Scott's example below, only something like ~ the first 18K
atoms have dihedrals, so you have no problems. Typically your protein or
dna or whatever that has dihedrals was at the beginning of the pdb, and has
the lowest atom id's possible. So the rule of thumb is that above 33,332
solute atoms, move to the "new" (amber 7) format prmtop.
Regards - Bob
----- Original Message -----
From: "Scott Brozell" <sbrozell_at_scripps.edu>
To: <amber_at_scripps.edu>
Sent: Thursday, October 09, 2003 5:30 PM
Subject: Re: AMBER: Compiling AMBER, ie, number of atoms limit

> Hello,
>
> There may be several lurking problems with large systems.
> In Amber 7 the NATOM format is I6 (it's I5 in Amber 6),
> suggesting a limit of 999999. But in recent threads
> Bob Duke and Tom Darden indicated that troubles may occur
> at substantially fewer atoms, perhaps > 333333.
> You can find those Reflectors via google searches:
> more than 100,000 atoms site:structbio.vanderbilt.edu
> atom number mismatch in coord and topology site:structbio.vanderbilt.edu
>
> On the other hand Amber 7 is known to work with,
> see benchmarks/rt_polymerase:
> NATOM = 141154 NRES = 42245
> Number of triangulated 3-point waters found: 41189
>
> Scott
>
> On Thu, 9 Oct 2003, franck wrote:
>
> > Dear Amber users,
> >
> > I would like to re-compile AMBER in order to deal with a higher number
> > of atoms (MD with SANDER). Increasing the number of atoms will increase
> > too the number of interaction.. So I imagine that I have to change, not
> > only NATOM parameter but some others. Could anyone tell me which are
the
> > parameters to check and to change for this purpose ?
> >
> > I thank you for your help,
> >
> > Chevalier Franck
> >
> >
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu