Wednesday, April 15, 2009

I am trying to use GROMACS (gmx) to do some molecular dynamics simulations on carbon nanotubes. Since this is not what gmx is built for, I need to do a bunch of things to get it to simulate them. This is also, pretty much, my first time using gmx. Please note that there are a few missing pieces with the following, in that it WON'T allow you to simulate carbon nanotubes but instead create a pretty tube that dissociates/was never together in the first place/blows up. I will write a full, accurate, tutorial once all the kinks have been worked out. I am posting this here in the hope of finding some help working out those kinks, or maybe giving a couple of chuckles to those that have this down cold <-I don't think there are many though.


1. got a pdb file of a small carbon nanotube and made each atom a residue with a unique name, for example:
ATOM 4 C NT1 1 2.624 3.131 -6.155 1.00 0.00
ATOM 5 C NT2 1 2.043 3.538 -7.386 1.00 0.00
ATOM 6 C NT3 1 0.707 4.023 -7.386 1.00 0.00

Here is the pdb file of the test specimen. This is something I will automate if it turns out there's no better way. I'll post the script too.

2. added carbon to encadv's atom type file, ffencadv.ATP

C 12.011 ; cnt carbon

3. added the following to encadv's include topology file, ffencadvbon.ITP I got this information from Christopher Stiles' site. It is rare to find such good instructions for something like this. However these are for a previous version of gmx, and you quickly run into problems when trying to follow it for 4.0. This is what set me on my own course to trying to solve this problem; without Christopher's site though, I would not have even known where to begin with tubes.

[ bondtypes ]
; i j func b0 kb
C C 1 0.14210 478900

[ angletypes ]
; i j k func th0 cth
C C C 1 120.000 397.480

[dihedraltypes ]
; Encad proper dihedrals. Note that there is a difference pf
; sign in the cosine definition between Gromacs and Encad. All angle values
; here are thus offset 180 degrees.
; i l func q0 cq
C C C C 1 0.000 167.360 1

4. I added carbon to ffencadvnb.ITP. Right now I don't have a reason why I added it, and if these are non-bonded parameters, then I'm thinking it shouldn't be there?

name at.num mass charge ptype sigma epsilon
C 6 12.011 0.0 A 3.84423e-01 3.08863e-01

5. Created ffencadv.N2T using information from Christopher Stiles' site

6. Updated encadv's residue topology database by adding a residue for each atom in the tube. This will be a problem for large tubes, but I can automate the process too unless better ways are to be found. Here is an example for the first residue:

[ NT1 ]
C C 0 0

C C 0 0

[ bonds ]
C C 0.14210 478900

C C C 120.000 397.480

Here is the file ffencadv.RTP

Ok I think that is all the changes to gmx files I made.


1. fired up gmx in cygwin and from the prompt:

pdb2gmx -f swnt.pdb -o swnt.gro -p

The console print can be seen here. Notice all the warnings about the short bonds? I originally thought these were from shifting the columns in the pdb file when I renamed the residues, but I realigned them with a PDB spec and still get the warnings. I think whatever is causing this could be a major problem.

swnt.gro posre.itp

It looks like whatever is wrong is most likely in the above. I'm going to leave it there for now until I know all that's correct.

No comments: