Same Protein Across Different Species In Pdb (Suitability For Docking?)

Question

[Santi]

Hi all,

Is there any reliable way of, given a PDBID, finding roughly all the experimental structures corresponding to the same protein (structure and perhaps function) in different organisms? For example,1A4H and 1UY6 are both HSP90 realizations in yeast and human. I guess I can rephrase the question like this: how can I find roughly all PDBs that are orthologues to a given PDB.

I know that "same protein", as I'm using it here, is quite an abusive and vague term. To make things clearer, our research is about docking and so here is a second related question: are there any rules of thumb as for when is it licit/useful to use structures from different organisms to perform docking studies?

Thanks a million

Answer 1

[Khader Shameer]

Definition of orthology is more specific for sequence data than structur data. For structural homologs you may consider structural members from same class, fold, superfamily, family or domain.

Since protein structure are more conserved than sequence (Read this seminal work for a background), for structural homologs I would recommend to take a look at protein structure classification databases like SCOP, CATH etc and use members from same family or superfamily for your structure analysis.

You may also consult structural alignment databases like PASS2 for selecting structurally similar proteins based on structural parameters like RMSD which is based on superposition of two proteins than mere sequence alignment.

Now let's take your specific example: 1A4H: This link provides classification information for this protein from SCOP, CATH, Pfam and GO. Practically you can select proteins with same annotations from any of these classes. But I would recommend you to use structural domain or family level information to get your structural homologs. Depending on your requeirement you can select your proteins, for example:

• Click here for all proteins in PDB with HSP90, you can filter this list further based on your requirement.

Answer 2

[Reyhaneh]

Since I am working on similar topics I have found a useful cluster of PDB files. " ftp://resources.rcsb.org/sequence/clusters/ contains the results of the weekly clustering of protein chains in the PDB generated by blastclust". This can be helpful in finding proteins which have similarity at sequence level and the structures exist in PDB as well. this is mainly similarity at sequence level. But if you want similarity from different aspects Pfam has a nice table for every family with PDB codes (e.g. http://pfam.sanger.ac.uk/family/PF00057#tabview=tab9). You need to generate this information for the PDB files you need or write a script to generate all of them.