[gmx-users] Handling PBC for Free energy calculations in plumed patched Gromacs

[Abhinandan Panigrahi]

Dear all,

I am trying to study free energy profiles for base flipping in a
DNA+Protein system via
umbrella sampling calculations on Gromacs 2016.3 patched with PLUMED 2.4.1.

On visualizing the GRO file which comes after equilibriation (and which is
used to make
the initial TPR that each window uses), I saw that part of the DNA molecule
has moved out
of the PBC box. The atoms in the DNA concerning my collective variable are
also coming out
of the PBC box (my system is huge, and the water box is not too big).

Simulations run using TPR generated from this GRO file have resulted in
profiles that are
most likely, incorrect (scatter data from production runs is far from the
minima obtained
in the PMF).

Q1) To make sure that the PLUMED calculations start with the entire system
inside the PBC
box, should I apply the trjconv command to the equilibriation GRO file, and
use `-pbc mol`
and `-pbc nojump` in succession to bring the part of the DNA that went
outside the box
back inside it. This file will then be used to generate the initial tpr.

Q2) In the PLUMED config file where I apply my restraints, I use the command
`WHOLEMOLECULE ENTITY0:<range of all dna+protein atoms>`, which as per my
understanding,
handles PBC for PLUMED calculations of the CVs. Does this conflict with the
PBC treatment
that GROMACS uses?

My CVs are angular in nature (bias applied over 2 dihedrals). The protocol
for umbrella
sampling was as follows:

- equilibriate system well (total 5ns of nvt + npt runs)
- use the gro file generated above to make initial tpr, which is used by
each umbrella
sampling window
- restraints/bias is defined in plumed data files passed to mdrun with
-plumed flag
- run wham on plumed generated colvars to build pmf

Thanks in advance,

-- 
Abhinandan Panigrahi

"To make an apple pie from scratch, you must invent the universe."