[gmx-users] getting a good .gro file from a pdb file while the topology is ready

[Mark Abraham]

Allen Smith wrote:

>>>> which I truncated to make a smaller bilayer, and added some more water
>>>> molecules, using programs other than gromacs. The final output I have is a
>>>> pdb file with complete solvation, and a topology file which I obtained from
>>>> the original top file by simply changing the number of residues of each
>>>> group in the last section.
>>> Umm... the atom numbers (not just the numbers of residues) are going to need
>>> to correspond between the .top and .gro files.
>> I think the OP meant changing the number of molecules in the [molecules] 
>> section of the .top file. I expect that within each molecule in the 
>> structure file the order of the atoms must agree between with the 
>> topology file. Certainly the order of the molecules in the [molecules] 
>> section must correspond to that in the structure file.
>>
>> However, the atom numbers cannot (in general), match in order between 
>> structure and topology file for any system with more than one copy of a 
>> molecule (e.g. solvent). For simple solvated systems where one gets 
>> focussed on the solute, it can look like correspondence is occurring and 
>> thus might be necessary :-)
> 
> However, in this case, the original poster has not only edited down the
> solvent, but the bilayer. 

The bilayer is just another set of molecules, like the solvent. There's 
nothing magical. As above, assuming the OP mis-used "residues" for 
"molecules" then they're on the right track for this part of the problem.

> I am reasonably certain that if I put in more than
> one copy of something which is in a .rtp file and _not_ in, say, ions.itp,
> then it will be in both times in the .top file

I think you are missing a point or two in comprehension here. The .rtp 
file is a database used only by pdb2gmx to help construct a 
correctly-connected topology for the residues in that database. I don't 
know about the behaviour of pdb2gmx on a structure file with multiple 
copies of the same molecule in different chains, but if it works, I 
expect it won't notice the same molecule is the product, and would thus 
produce two differently-named [ molecule ] sections for a dimeric 
system. In any case, the OP did not need to have been using pdb2gmx, and 
so the contents of the .rtp file are a red herring.

> (after all, two
> different-position glycines are in twice in the .top file...).

pdb2gmx has atoms and residues as its two levels of structure. The 
topology file has atoms and molecules as its two levels of structure. In 
the latter, residue names and numbers are retained purely for human 
convenience, say, in constructing index groups. Since these glycine 
residues are parts of molecules, they have to be replicated. This has 
nothing to do with replication of molecules, however. You could define 
multiple different [ molecule ] sections with different names for the 
same molecule and then have multiple entries in the [ molecules ] 
section, or you could have just one and have a single entry in the [ 
molecules ] section. This brings us back to my original point that in 
the latter case the numbering of the atoms in [ molecule ] sections does 
not need to correspond to the structure file.

Mark