[gmx-users] Advice on High Precision Structure File Formats

[Andrew DeYoung]

Hi,

If you have time, I would be grateful for any advice you have about which
structure file format that I should use.  I am attempting to create a carbon
nanotube.  However, the carbon-carbon bond length that I would like to use
is 0.1415 nm, which means that I would like a structure file format that can
take positions up to accuracies of at least 0.0001 nm.  

Unfortunately, the .gro file format only accepts positions up to accuracies
of 0.001 nm.  This means that if I use .gro, I will need to round the
carbon-carbon bond length to 0.142 nm or 0.141 nm, which I would prefer not
to do.  

Now, I know that the Gromacs code itself uses many more decimal places
internally for running MD (single precision give a lot of decimal places and
double precision gives even more).  I think this means that if I specify the
carbon-carbon bond length in my topology as 0.1415 nm in [ bonds ] (and the
.itp file format, and others, allow this type of precision) then even if I
must use 0.142 nm in my starting configuration .gro file, the bond lengths
will equilibrate to ~0.1415 nm if I run a short simulation (given that the
Gromacs code internally uses high precision).

Yet, what if I need my starting configuration to have high precision bond
lengths (0.1415 nm)?  I plan to freeze the carbon atoms using a freeze
group, and it would be easiest if I could do this from the get-go (i.e.,
without running equilibration in which the carbon atoms are unfrozen).

Do you have any advice about which file format I should use?  I need not be
"constrained" by .gro
(http://www.gromacs.org/Documentation/File_Formats/Coordinate_File#On_the_ne
ed_for_a_.gro_file).

.pdb, it seems
(http://en.wikipedia.org/wiki/Protein_Data_Bank_(file_format)#Example),
gives position precision up to 0.001 ANGSTROM (or 0.0001 nm), which means
.pdb will give me just enough precision.  Or should I instead write my
starting configuration in a .g96 file, which allows very high position
precision?  However, other than
http://manual.gromacs.org/current/online/g96.html, I am having some
difficulty (even with Google) locating detailed information about the .g96
file format.  Unfortunately, my institution I do not think has a GROMOS-96
license, so I cannot access its manual.

Between .pdb and .g96, which would you recommend?  Or is there any other
structure file format that I am not thinking of?

Thanks so much.

Andrew DeYoung
Carnegie Mellon University