Re: [AMBER] Fwd: q4md FF with two "small" molecules

From: Jason Swails <jason.swails.gmail.com>
Date: Mon, 21 Sep 2015 07:23:50 -0400

On Mon, Sep 21, 2015 at 6:11 AM, Francesco Pietra <chiendarret.gmail.com>
wrote:

> The outcome of MD was most disappointing, and a remedy applied most
> intriguing.
>
> The largest ligand had been optimized, and no negative frequencies arose
> (Gaussian at the computer center). The gaussian log gave the RESP partial
> charges and mol2 on the PyRED server. Minimization (NAMD 2.10, water box,
> neutralized), even at timestep 0.001, caused a vast deformation of one of
> the three cycles (the one bearing N-CO-NH that I discussed before), and of
> the C-O-P angle (phosphate phosphorus). Gradual heating to 300k with
> timestep 0.001 did not correct the structure. Very good parmchk2 score
> notwithstanding, I had to conclude that the parameterization by analogy
> (antechamber parmchk2) was no good.
>

​This sounds like a misassignment of atom types to me. A region that
*should* be planar but gets deformed would result from assigning atoms that
*should* be aromatic to atom types that are not; this is what I suspect may
have happened (you should check this carefully in your generated mol2 file).

If this is what happened, then database scores can still be very good,
since it finds good matches with the assigned atom types -- it is simply
the assigned atom types that are wrong. You can try to fix these by hand
in the mol2 file (and run it through parmchk2 again) and see if you get
more reasonable behavior.

For a problem this severe, I would expect atom type misassignments to be
fairly obvious.

HTH,
Jason

-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Sep 21 2015 - 04:30:05 PDT
Custom Search