Protein interaction experimental pdb data

Question

[Bioaln]

Hello. I've been recently trying to obtain experimental data on protein interactions, up to the pdb entry. I've come to realize, there might not be such an easy way of obtaining this data - For example Stringdb or intAct only provide Uniprot or some other accession entries (If I'm not mistaken), And obtaining exact pdb from just the Uniprot is currently not possible - at least not precisely.

Is there any other way of obtaining experimental PPI pdb data?

Thank you very much in advance.

Answer 1

[Jean-Karim Heriche]

I am not sure what you mean by not possible to get exact PDB from UniProt. PDB is searchable by Uniprot entry. UniProt has a PDB cross-reference file and PDB also provides a list of structures to Uniprot mapping.

Also Andrew Martin's group at UCL provides a web interface to the PDB/UniProt mapping with REST access.

You may also be interested in the EBI's SIFT (Structure integration with function, taxonomy and sequence) resource.

[Bioaln]

Thanks for the answer, I obviously wasn't clear enough. I knew how to get from unuprot to myltiple pdbs connected with it. I'm tying to obtain exact pdbs with their locations - as in interaction only a small part of pdb is involved. Stringdb doean't provide this detailed data as far as I know.

[Jean-Karim Heriche]

I still don't get what you mean by "exact pdbs with their locations". Do you mean the contact site between interacting proteins ? This should be in the PDB records and in SIFTS. Or do you mean finding out which PDB entries corresponds to protein-protein interactions ? For this, you could go through all PDB entries and extract those that have more than one protein.

[Bioaln]

Yes, the contact site with corresponding pdbs is in my interest. I admit I have not yet checked sift, thanks for that suggestion. I didn' know multuple uniprots can share same pdb ID? So you are saying that two proteins sharing same pdb ID are necessary involved in an interaction? If this is true, you helped me very much indeed.

[Jean-Karim Heriche]

Yes a PDB entry corresponding to the structure of two interacting proteins should reference two UniProt accessions e.g. 4AF3. For more than two interacting proteins look for more than two UniProt accessions. Depending on your end goal, you may have to deal with redundancy.

[Bioaln]

Thank you very much, redundancy is at this point inevitable indeed, but it is a start.

[Bioaln]

I've forgotten to ask you another question regarding this matter, thank you in advance for any reply. There are may PPI databases (intActmStringdb...), in all of them one can obtain interacting partners but not the uniprot location of this interaction. Is there any way of obtaining this information - so to explain once again I apart from interacting accessions want to obtain uniprot location of the interaction.

[Jean-Karim Heriche]

Most databases allow you to retrieve a UniProt accession for interacting partners. You can then use these to look up structural info from either PDB or UniProt. See for example the structure section of this UniProt entry:
http://www.uniprot.org/uniprot/Q9NQS7?conversationContext=3#structure

[Bioaln]

Okay, so there is not a direct mapping already made. Even from Uniprot site, the interaction section contains databases which don't provide exact loci of interaction site - there are more or less computationally predicted interactions, with no specified loci.