[gmx-users] Expanding a .top file to have all connection information

[Mark Abraham]

On Mon, May 20, 2013 at 4:36 AM, David Rosenman <davidjrosenman at gmail.com>wrote:

> Hello again,
>
> Sorry for the delay in my response.
>
> The original top file is:
> http://wikisend.com/download/951610/APPTM_WT.top
>
> The grompp -pp processed top is:
> http://wikisend.com/download/851584/APPTM_WT_processed.top
>
>
> The connectivities that are not explicitly mentioned in either of these
> files are the connections between the second molecule of the homodimer,
> between all DPC detergent molecules, and between all water molecules.
>

The inter-molecular interactions of all the moleculetypes *are* present in
the preprocessed file. Whether some third-party script has enough tech to
read them all is another matter :-)


> GROMACS already knows what the connectivities of these are because it can
> interpret both included itp files and multiple molecules. But, top2psf
> doesn't.
>

Right. As you said earlier:

> All of the include records will be ignored, as will multiple molecules.

The first problem is solved, but apparently top2psf was never designed to
do the latter. That's a very common case - busy people volunteer to write
just enough functionality to make what they need work and contribute it in
the hope that it is useful to someone else. If you need it to do something
else, then you can extend the functionality (and preferably contribute it
back!)


>> Or just do it by hand and replace the lines in the .top with each
> protein chains .itp file.
>
> This won't work because the generalized atom indices in the itp files don't
> match the actual atom indices in the structure for each molecule I need
> included.  And there are many molecules to deal with here. . .ie, 100000
> water molecules! I suppose one could write a script to make the
> translation, but it sounds like it may be a challenging task.
>

Right. Differences in file formats' expectations about (over-)specifying
repetitive information is one of the things that makes conversion utilities
difficult to write. The [moleculetype] functionality is intended to
facilitate re-use, but if psf does not have a similar functionality, then
you will have problems. If it does have that functionality, then you can
side-step the limitation in psf2top you describe (of only processing one
[moleculetype]) by feeding subsets of the (post-processed) .top to top2psf
and doing the cut and paste in psf afterwards to put it together. But this
is starting to get off-topic for a GROMACS mailing list discussion ;-)

I expect the usual solution here is to use the psf generation tools to
build the equivalent topology. There are probably gotchas with atom naming
and ordering that might be problematic if you do it this way and expect to
reuse a set of coordinates.

Mark