Fwd: Atoms too fast/periodic cell too small with ABF protein-ligand

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Mon Jun 26 2017 - 05:28:49 CDT

I must add that the attempted ABF was "Conformation:Bound". For rmsd for
the ligand I had chosen all heavy atoms.
By choosing all atoms of the ligand, except hydrogens at the methyl groups,
the simulation also crashed at the first step, this time for two H atoms of
the protein, one (HG1) very far from the ligand, the other one (HA) close
to the ligand.

With "Conformation:Unbound" the ABF went to completion without errors.

Hope this helps suggesting what to do.

As I said, there was no problem with MD equilibration for this system, at
the same ts=1.0fs, along a trajectory of during 58.2ns.


---------- Forwarded message ----------
From: Francesco Pietra <chiendarret_at_gmail.com>
Date: Sun, Jun 25, 2017 at 12:56 PM
Subject: Atoms too fast/periodic cell too small with ABF protein-ligand
To: NAMD <namd-l_at_ks.uiuc.edu>


I am attempting a protein-ligand ABF, following the 2017 tutorial, by using
a 58.2ns problemless equilibrated system (ts=1.0 fs, bond restriction on
water only) in a TIP3P water box on a main pure-CPU cluster. Rather large
protein, organic ligand as accurately parameterized as I could, by fitting
torsions and water interaction.

I am experiencing immediate "atoms moving too fast" (two H atoms of the
ligand) when using a linux 4-core cpu desktop, or "periodic cell has become
too small" on a linux GPU workstation. At the moment I have no access to
the cluster.

I used ts=1.0 fs, i.e. no bond restriction, except for TIP3P water, as in
the equilibration.

As a possible cause, that I was unable to verify, is the setting of Euler
and polar angle in the colvars definition. That is , I used the following
values from "cv printframe"

>Main< (Conformation:5syf-SAA1) 28 % cv printframe
> 0 1.88926419197828e+01 5.22845307973583e+01
> -3.39612142392650e+01 9.26730407864645e+00 1.01135055762437e+02
> -5.05311208874754e+01 0.00000000000000e+00 0.00000000000000e+00

by discarding the first two, and using the directly measured intercenter
distance of 27.6A in place of 18.89 from printframe, i.e., as follows:

harmonic {
   colvars r
   forceConstant 0.0
   centers 27.6 # OK measured

harmonic {
   colvars Theta
   forceConstant 0.0
   centers 52.3 # from printframe

harmonic {
   colvars Phi
   forceConstant 0.0
   centers -40.0 # from printframe

harmonic {
   colvars Psi
   forceConstant 0.0
   centers 9.27 # from printframe

harmonic {
   colvars theta
   forceConstant 0.0
   centers 101.1 # from printframe

harmonic {
   colvars phi
   forceConstant 0.0
   centers -50.5 # from printframe

Should this assignment of colvars be correct, where to look for the causes
of the instability of the system?
I must confess that I am the first time at such an ABF beyond the tutorial

Thanks for advice

francesco pietra

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:23 CST