[gmx-users] pdb2gmx for non-proteins

[Anton Feenstra]

Christoph Freudenberger wrote:
> Hi Anton,
> 
> Anton Feenstra wrote:
> 
>> Christoph Freudenberger wrote:
>>
>>> Hi there,
>>>
>>> I'm thinking about how to generate the top's for tripodal
>>> scaffold-molecules bearing different substituents:
>>>
>>>     R1
>>>       \
>>>        Scaf--R3
>>>       /
>>>     R2
>>>
>>> The scaffold is an aryl-ring and R1-3 should be arbituary
>>> substiutents.
>>> I wonder if it is possible to add rtp-entries for Scaf and
>>> the R's and let pdb2gmx do the job.
>>> Is pdb2gmx able to handle non-chain-like molecules?
>>
>>
>>
>> No and yes. Initially, bonds are only made between consecutive
>> residues (as I describe below). Then, there are the 'special
>> bonds' from the file 'specbonds.dat', where additional bonds
>> are defined. So, you can have bonds genereated 'sequentially'
>> e.g. between R1, Scaf and R2, and have a 'specbond' between
>> Scaf and R3. Note, that the sequential bonds are always made,
>> they are defined in the .rtp, but the specbonds are only
>> made if a distance criterium is met (the reference distance
>> from 'specbond.dat' +/- 10%). You will still need an .rtp
>> entry for R3, of course. (Compare e.g. Cys-Heme and Cys-Cys
>> bonds.)
>>
>>> How are the links recognized? Automaticly (e.g. by distance)
>>> or do I have to define them?
>>
>>
>>
>> Automatically, from the definitions in the .rtp by making bonds
>> to, e.g., atom '-C' ('C' atom in previous residue), or '+N' ('N'
>> atom in next residue). Distances are not checked here.
> 
> 
> If understood that correctly, pdb2gmx would look for something
> like this:
> [R1-N]-[C-Scaf-N]-[C-R2]

No, the 'N' and 'C' are only examples. If you look in the .rtp
files, in any of the amino acids, you
will find interactions defined like:
[bonds]
-C	N
[angles]
-C	N	CA

Where '-' refers to an atom in the previous residue.
So, in your case you'd want for R1 a bond from 'N' to '+C',
and for R2 and bond from '-C' to 'N'. (Of course, both C's
need to have different names in reality.) Similar thing
for angles and dihedrals.
That really only leaves the 'Scaf-R3' bond for 'specbonds'.

Now I come to think about it, I remember looking into this
part of the code a while back, and it might be that there is
no restriction on how far 'back' or 'forward' pdb2gmx will
search to find the matching atom name. If that is the case,
you can have a .pdb file with R1, Scaf, R2 and R3, and if
atoms 'C1', 'C2' and 'C3' in Scaf link to R1, R2 and R3
respectively, and there is no 'C3' in R2, then R3 will
be linked to 'C3' in Scaf. Not sure about this, and it is
not too easy to find exactly where in the code this happens
(topology parsing is *not* trivial to code!), but you could
try. I'd have no objections to having this limit removed,
if it would be there, but there should probably then be
a warning about creating bonds across residues.

> But of course my system goes:
> [R1-N]-[C-Scaf-C]-[N-R2]
> 
> I suppose this is only possible by defining all three bonds
> via specbonds.dat.
> I'll give it a try.
> 
> thanks


-- 
Groetjes,

Anton
  _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610           |
|             | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/       |
|             | "If You See Me Getting High, Knock Me Down"           |
|             | (Red Hot Chili Peppers)                               |
|_____________|_______________________________________________________|