AMBER Archive (2007)

Subject: Re: AMBER: MD simulations with Pt atom - how to keep square planar geometry?

From: gsciaini_at_qi.fcen.uba.ar
Date: Thu Nov 08 2007 - 10:09:31 CST


Pablo,
Try adding virtual atoms and applying shake...that could keep the right
complex structure...might work...By the way, how are tango classes going?

Cheers!
G

> I'm trying to run MD simulations on a system comprised of a DNA
> molecule and a platinum complex, and for that I developed ad-hoc
> parameters for the Pt atom by fitting energy values from DFT
> (B3LYP/LACV3P**++). I added all the parameters on an frcmod file read
> by leap and I'm able to _run_ minimizations and MDs with sander,
> although not as satisifactorily as I wish to.
>
> The _only_ problem is that the square planar geometry around the Pt
> atom is not stable, and the ligands bend towards a tetrahedral-like
> geometry (whether minimizing or running an MD simulation).
>
> I have 2 pointers to this issue:
> 1) when trying to fit the DFT energies to the MM, it was found that
> the 90 degree angles for the ligands were enough to describe the
> variance in energy, so the 180 degree N-Pt-N angles have a constant of
> 0
> 2) when trying to fit an oop/improper torsion parameter for the Pt
> atom, I found that the variation in MM energies (due to increases in
> bond distances and angles) was much larger than the one observed in
> DFT, so I (initially) set it to 0.
>
> I tried 2 ways to fix this issue:
> 1) setting a non-zero Pt oop (actually, 2 sets of oop's, involving the
> Pt atom and 2 sets of 3 N atoms bound to it). It appears that leap
> does not use it in any case, because the central atom is not a
> tricoordinate atom, and it doesn't appear used in the parm file
> generated by leap (nodummy_500.top). After an md simulation in vacuo,
> the Pt is still bent.
> 2) using a pair of dummy atoms bound to the Pt in the axial positions
> of the square, with angles at 90 degrees of each of the N substituents
> with a high K; a 1A bond distance with a high K; torsions specified
> but at 0; and 0 vdW parameters and mass. After an md simulation in
> vacuo, the Pt is planar, but the dummy atoms are at 33A from the Pt,
> and the simulation crashes with a SANDER BOMB (and the bond energy
> goes to the roof).
>
> I'm attaching the files I'm talking about (frcmod; .crd and .top for
> both dummy and non-dummy test cases; sander input file as md_rest.in).
>
> I'll appreciate any suggestions as to what can I do to keep the Pt in
> a square planar geometry...
>
> Thanks for your time in advance!
> --
> Pablo Englebienne, PhD Student, Moitessier research group
> Department of Chemistry, McGill University
> 801 Sherbrooke Street West
> Otto Maass bldg, room 206
> H3A 2K6 Montréal, Québec, Canada
> Tel (514) 398-5501 Fax (514) 398-2382
>
> "Progress is made by lazy men looking for easier ways to do things." -
> Robert A. Heinlein
>

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