On 6/9/2012 1:29 AM, Francesco Pietra wrote:
> Hi:
> Adding ions to a regular protein-single-residue-small-molecule ligand
> of ca 260 residues (built through autodock from previously
> equilibrated protein, a procedure familiar to me, normally without
> problems) results in the protein not being centered in the TIP3P box.
> This is caused by a ball-like group of 27 Cl- ions, with one Na+ at
> the center, lying outside the protein surrounded homogeneously by Cl-
> and Na+. The aim was to get a ca 0.6 M NaCl concentration
>
> Commands executed with ambertools12 LEaP:
>
> -- xleap ....leaprc.ff12SB
>
> -- source leaprc.gaff
>
> -- prep and params for frcmod.ionsjc_tip3p, calcium++ (6 present,
> bound to the protein, and retaining their position on protein
> equilibration), and ligand.
>
> -- loadpdb (two identical models).
>
> -- solvate box model1 TIP3BOX3P 12 0.85 (added 23545 waters;
> dimensions 107, 118, 69; vol 876953 A^3).
>
> -- solvateoct model2 TIP3PBOX 12 0.85 (added 39438 waters; vol 1315698 A^3).
>
> -- for both models: neutralize (addions Na+ 0, as the protein complex
> has charge -27.00000), then "addions Na+ 200" "addions Cl 200".
>
> -- saveamberparm ...
>
>
> -The ligand, treated by antechamber, had RESP charges.
>
> I carried out the above with either the protein retaining its crystal
> water, or not. Same problem. I repeated everything from scratch with
> newly equilibrated protein, with same problems.
> Minimization/equilibration does not correct; actually it emphasizes
> the problem, making one side of the box even more disequilibated.
>
> I wonder why those misplaced Cl- (and one Na+) have replaced water
> molecules inhomogeneously around the protein. Was the number of added
> ions too large for the water molecules present?
When I wrote the addIons code, I didn't think to test such high
concentrations. I know of one other bug in addIons that is easier to
test for and is relatively innocuous: adding 2 Cl- to an Na+ does not
give a linear arrangement.
The code uses an octree data structure (each cube divides into 8 smaller
cubes recursively) to optimize space; it is possible the bug is due to
an error in pointer arithmetic for the data structures involved. Using
memory-checking instrumentation like Purify might help.
It may be possible to work around it by adding smaller amounts of ions
with each command more times, or more simply by following the less
recommended approach of addions before solvate (creates more vdw voids
by placing ion then removing potentially a few waters).
Bill
>
> thanks for advice
>
> francesco pietra
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jun 09 2012 - 10:00:03 PDT
