Retrieving Protein Structures

Question

[Pals]

I want to make structural analysis of lysosomal proteins found in mammals. To make start, I would like to retrieve as many structures as possible from the biological database (PDB). These structures show dynamic structural features (many fold types, variation in size and multiple cellular location). I have tried two general strategies - search by keywork "lysosomal" and get results based on deposition date of structure. I also found an article on proteomic analysis for lysosomal proteins where the author had provided GI for each proteins. In this case, I made protein blast against PDB and selected the lysosomal proteins. Both ways were quite time consuming and again I think my list is still incomplete. Are there any databases for these kind of structures? I would be pleased to get relevant suggestions and help regarding my problem.

Answer 1

[Neilfws]

You can do this search directly using the advanced search link at the PDB website.

1. In the query drop-down, scroll down to "Biology->Cell Component Browser (GO)"
2. Enter lysosome: the auto-suggest will provide "lysosome (GO ID:5764)"
3. Click "find in tree", then click the link for "lysosome (GO ID:5764)"

Voilà, 330 structures. Add other qualifiers to your search if you wish and submit query to see/retrieve the structures.

[Chris Evelo]

Wonderful! Thanks!

[Pals]

Thanks a lot. This is really simple but of course, I would need to take the one with longest sequence which would be trivial using UniProt sequence ID and ProteinBlast.

[Neilfws]

I think it would be even more trivial to retrieve chain sequences from the PDB and select the longest one.

[Pals]

I found out that the most easiest approach would be UniProt itself. I searched proteins with keyword lysosome and the options for result was divided into different categories and I was able to get result with 3D structures only. Here the result was lowered by 50% (164)

Answer 2

[Chris Evelo]

What you might want to try is searching [?]UniProt[?] instead. That will allow much more structured search strategies and it has cross-links to [?]PDB[?] for the relevant structures. Now the tricky word in that last sentence is of course "relevant". There definitely are a lot of structures in PDB that are not cross-linked from UniProt, especially many structures for proteins based on (sometimes deliberate) mutations. If you are also looking for those the UniProt approach will not work.

If you Google for "lysosomal proteins database" you will find that there are quite some databases for lysosomal proteins identified in 2D-gel proteomics approaches. I haven't tried these, but it is likely that most of them contain links to UniProt that you could again use to find PDB entries, or maybe even direct PDB links.

[Chris Evelo]

It probably will. But a more structured approach would be to use the Gene Ontology category "lysosome": http://amigo.geneontology.org/cgi-bin/amigo/term-assoc.cgi?gptype=all&speciesdb=all&taxid=9606&evcode=all&term_assocs=all&term=GO%3A0005764 and its children. You could either take them from GO directly (the link will show the human ones) or you could search UniProt for all proteins with these associated GO terms, which is probably more effective. In this way it is 3 steps (GO lysozome -> UniProt protein -> PDB structure) but that makes sense.

[Pals]

Thanks a lot for the suggestion. In that case, would you think the search with the keyword "lysosomal" would suffice to get maximum of the related structures? Actually I had used Uniprot as secondary database for information retrieval for PDB structures. The aim of the analysis is just looking overall structural features by categorizing the structures into different fold types, inspecting the catalytic sites, surface charge etc. I think UniProt approach works in this case.

[Chris Evelo]

It probably will. But a more structured approach would be to use the Gene Ontology category "lysosome": http://amigo.geneontology.org/cgi-bin/amigo/term-assoc.cgi?gptype=all&speciesdb=all&taxid=9606&evcode=all&term_assocs=all&term=GO%3A0005764&session_id=3107amigo1301220458&action=filter and its children. You could either take them from GO directly (the link will show the human ones) or you could search UniProt for all proteins with these associated GO terms, which is probably more effective. In this way it is 3 steps (GO lysozome -> UniProt protein -> PDB structure) but that makes sense.