Getting ready to run

• A python interpreter (python.org) and a copy of LAMMPS
  (LAMMPS Molecular Dynamics Simulator: http://lammps.sandia.gov/)
• The code files included in this package – you will want to replace some of these default files with your own lipid files depending on your own desires and preferences. The included files will work with the force-field of Orsi et al, but you may use whatever force field is appropriate to your systems. The current example uses a DOPC molecule (Orsi et al) as an example, hence the use of the suffix “dopc”, but you can use any lipid scheme of your choosing.

Adding pair styles and dipole bending.

• The example lipids given rely on point dipole routines and these can be downloaded from here and included into your build of LAMMPS before compilation to ensure that these can be run. We are working on a more ‘Vanilla’ lipid example that you can use right out of the box with recourse to downloading and installing.
• More information on the point dipole schemes employed is found in…

Notes on numbering (atoms and molecules)

• All numbering should be contiguous where possible. We cannot test for every variation, so with your numbering scheme for atoms and molecules please ensure where possible numbers run contiguously.
• We start molecule numbering from zero (0) and atoms from one (1) and this is partly historical and we suggest that you do the same to ensure you don’t run into obvious problems.

Letting Python do the heavy lifting

• Get a single molecule system working. It is imperative that a single lipid system works on its own since systems will be built on these foundations. Ensure you have a system with no errors that runs with LAMMPS with no errors. If you are using the existing molecule representations (included), make sure that you have the requisite files and pair styles. See here for copying and making the pair-styles in your LAMMPS build.
• Use the meta.* and forcefield.* files to build the parameters of the systems – the LJ interactions and scaling between cross terms. The Lorentz-Berthelot scheme is implied, but this can be changed by adding a row in the Scaling section to correspond to force an extra scaling term between a pair of atom types.
• Set the dimensions of the simulation environment in and the number of solvent molecules/
  lipid molecules. Check the filenames in the cfg.py correspond to filenames in where appropriate. If you wish to use a scaled nearest neighbour bonding scheme based on sigma values of individual atoms then set the value in this file.

Each integer value represents the number of each molecule to be represented in bilayer form in the system. So for a two phase system comprising 128 molecules in total the line molecules = [64, 64] will specify a bilayer with 50:50 composition of both those phases. Please make sure that the number of lipids on each leaflet is a square number. 64, 81, 100, 121, 144…etc which would mean that in total there are relatively 128, 162, 200, 242, 288 lipids in total for example. No checks will be performed to ensure this arrangement.

The arrangement of the phases is controllable with the and commands which dictate how the molecules are arranged. will attempt to place each phase alternately into the bilayer and will randomize this output to give a randomized initial condition. Switching these commands on and off is performed with the use of the “True” boolean.

• data.* files comprise the structural information of the lipid and water molecule. Ensure that positions of the lipid are set according to the centre of the system volume but the lipid head is off-set by a positive amount such that the middle of the bilayer corresponds to positions at or close to 0.0.
• If you are good to run then simply run:

  python SmartBuilder.py

  This will rebuild the following files:
  forcefield.dopc_scaled
data.sys

  You can then run the LAMMPS simulation in the usual way i.e.:

  ./lmp_serial < in.dopc