0
$\begingroup$

For pdb files is there a simple programmatic way to determine if the file is made up of homodimers or heterodimers.

I have many files so an online tool or some psuedocode/scripts would help more than manually assessing each one.

$\endgroup$
  • $\begingroup$ Could you provide us with an example? Preferably one that is a homodimer and one that is a heterodimer. $\endgroup$ – terdon May 19 '15 at 15:21
  • 1
    $\begingroup$ I don't think this post deserves closing. I have tried to tidy up the question so its easier to understand. Correct me if I am wrong, but is OP asking if there is a programatic way to check if in a dimer both monomers are either the same or different? $\endgroup$ – James May 19 '15 at 19:17
  • $\begingroup$ Have you tried simply checking the sequences? same sequences for proteins = homodimer... I have distant memories of pdb fileformat, but as far as I remember individual protein chains are stored separately, but I may be dead wrong on this one. $\endgroup$ – Nandor Poka May 19 '15 at 20:44
  • $\begingroup$ How comfortable are you with scripting? $\endgroup$ – James May 28 '15 at 23:59
  • 1
    $\begingroup$ There was a recent discussion on this sort of question in meta. What you're asking here is not actually about biology, but about programming. Furthermore, as it stands this question could be interpreted as a "give me the codes" question. What have you tried so far? $\endgroup$ – James Jun 29 '15 at 17:17
1
$\begingroup$

I'd just compare the amino acid sequences of the individual protein chains. With some leeway for residues not observed in the density, ie only require 90 or 95% sequence identity, or ignore gaps.

This will not tell you anything about whether the dimer (if you observe several molecules in the asymmetric unit) is actually real. For that, you'd need to do an experiment, literature research or see whether software like the PISA-server could give you a clue.

| improve this answer | |
$\endgroup$
1
$\begingroup$

In case you want to write your own subroutine,
1. Scan the whole .pdb file for number of chains present, if multiple then proceed.
2. Create index_counter for each amino acids of a single chain.
3. Now, For SINGLE chain. While you scan for each amino acid perform a i=i+1 operation for that specific amino-acid_counter.
4. Compare the amino-acid_counter between/among the chains.
5. Exact match = homomer or non-exact match = heteromer.
Hope you get the logic.

| improve this answer | |
$\endgroup$
  • 2
    $\begingroup$ I think you will run into trouble with real live PDB-files here - often there are gaps in the sequence because they were not observed in the electron density, or the numbering is inconsistent. I would suggest to rather extract the sequence from the PDB-file and align them (eg using fasta). $\endgroup$ – Gerhard Jun 4 '15 at 7:21
  • $\begingroup$ @Gerhard yeah! but that depends on how good is the resolution and R-free value of your .pdb is. When you are talking about analyzing .pdb you are supposed to take a higher resolution <1.6A. This covers all the atoms vwd radii(except H), no atom-missing issue is possible. Literature says. $\endgroup$ – diffracteD Jun 4 '15 at 11:35
  • 1
    $\begingroup$ A) if you limit yourself to anything better than 1.6A, there are not many structures left, and (according to Murphy) most likely not the proteins you are interested in. B) whether you have gaps and missing parts in a crystallised molecule is not necessarily down to resolution and Rfree (although there is a correlation). Some parts of a molecule - in particular loops - are simply more flexible and hence disordered in the crystal - no resolution in the world will make these appear. C) Could you post the reference you mentioned - quite interesting as a crystallographer. $\endgroup$ – Gerhard Jun 4 '15 at 12:23
  • $\begingroup$ It's a general methodology as reported in every data-mining literature, that's why I emphasized on resolution. And as per my experience, I found that if you have a higher resolution, then random coils(most of the time) shows fluctuating occupancy, but still be present. Although you are right, some gaps stays sometimes(specially in multidomains proteins). However the subroutine can be configured to eliminate any structures with breaks and analyze further. But to reach publication quality I think high-res is always needed. - it's always nice to see a crystallographer around. $\endgroup$ – diffracteD Jun 4 '15 at 13:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.