[gmx-users] ligand-polymer interaction in triclinic unit cell

Wei-Ta Li weitali.md at gmail.com
Fri Feb 8 01:36:47 CET 2019

Hello everyone,

I am trying to simulate interaction of polymer with ligand. The polymer is
represented by infinite sheet build by propagation in 2 dimensions of
monomer obtained from crystal structure. Since the polymer was crystallized
in triclinic unit cell, the propagation of the sheet also appears to be
triclinic. The topology was obtained by "gmx x2top ... -pbc". The schematic
representation of the  structure before simulation is shown below where the
monomer units are *, L is the ligand, \ are boundaries of the unit cell and
water is omitted for the sake of simplicity:

\                \
 \                 \
  \                 \
   \                 \
    \       L         \
     \                  \
      \                  \

System was simulated with typical parameters and "pbc = xyz",
"periodic_molecules = yes" flags. After extraction of trajectory with "gmx
trjconv -s whatever.tpr -f whatever.trr -o whatever.pdb -pbc nojump -dt
10", the structure looks like scheme below:

\                \      **
 \                \      *
  \                \
   \                 \
    \       L        \
     \                 \
      \                  \
       \  ************\
        \ *************\

>From Gromacs manual I understand that such representation is expected
because regardless of the input box shape the simulation proceeds in the
brick-shape form: "Even when simulating using a triclinic box, GROMACS
always keeps the particles in a brick-shaped volume for efficiency, as
illustrated in Fig. 3.1 for a 2-dimensional system". Nonetheless I have 2

1) Why "-pbc nojump" processing does not result in the representation like
in Scheme_1? Shouldn't it put the atoms back in the original box like
stated in the description: "nojump checks if atoms jump across the box and
then puts them back"? Or does it mean brick-shaped box which Gromacs uses
for simulation? Is there other way to have representation after simulation
as in scheme 1?

2) Second question is somewhat naive, but nonetheless. Is such skewed
representation of the polymer sheet valid for simulation set up? In other
words, does this "skewness" have effect on polymer-ligand interaction? In
the brick shaped representation the left side has 3 layers, but the right
is gradual ladder-like increase of layers from 1 to 3, and then there is a
fragment of polymer on top of the box. Does the setting "periodic_molecules
= yes" take care of it?

Thanks in advance!
Wei-Ta Li

More information about the gromacs.org_gmx-users mailing list