It's true. It's all in the README file. Thanks Axel.
You can download the code you need for that example at:
I created a new pair style for the silly reason that there are many different particle types in the example, and many of them are purely repulsive. Right now LAMMPS does not really have an official pair style which supports automatic mixing of Lennard-Jones parameters for purely repulsive particles. Rather than introduce a new pair style, I'm happy to modify the existing "lj/cut" pair_style to support this, if Steve will accept the change. (About 8 lines of new code.)
Here are several more comments about this example:
1) The lipid mixture does not work very well. This is my fault. The original paper by Brannigan&Brown only describes one type of lipid (DPPC), and parameters were chosen accordingly. I deleberately added a second lipid type to demonstrate how to mix different molecules together using the moltemplate "new random" command. However this causes strange defects in the membrane over long timescales. If you don't need a mixture of lipids, just use DPPC and ignore DLPC.
2) the protein model is still a little too long to fit comfortably in the membrane, and it may tilt slightly during the simulation. Getting rid of the "DLPC" lipids will probably help, since they are shorter and they make the membrane thinner.
3) David Stelter and Saeed Momeni Bashusqeh have added the Cooke and Desrno lipid model into moltemplate (as well as well as many lipids from the MARTINI and SDK force fields into moltemplate), so there are many other membrane models you can choose from. (Assuming you don't want to invent your own... I need to remember to add the new lipids examples to the GitHub page. If I forget, please remind me.)
Glad to see people are using moltemplate!(especially for coarse grained biomolecules)