I was wondering how many human proteins have a solved 3D structure. Is there a database with only human proteins? I looked at pdb but couldn't find a filter.
$\begingroup$
$\endgroup$
6
asked Dec 23, 2011 at 4:27
-
$\begingroup$ I think you are going to have a lot of issues with redundancy, especially if you want to know the number of unique proteins. $\endgroup$– GWWCommented Dec 23, 2011 at 4:35
-
$\begingroup$ @GWW I guess that writing a small script can get rid of the redundancy. $\endgroup$– Gergana VandovaCommented Dec 23, 2011 at 4:36
-
1$\begingroup$ You may have better luck asking this question on biostar. $\endgroup$– GWWCommented Dec 23, 2011 at 5:00
-
2$\begingroup$ Notice that “solved” is very subjective. Not all of the structures are of high quality, and some of them are just plain incorrect due to experimental errors. $\endgroup$– Konrad RudolphCommented Dec 29, 2011 at 9:29
-
1$\begingroup$ This is a very subject question for two reasons: The Protein Databank accepts NMR and crystal structures of proteins which offer different degrees of resolution and accuracy. These structures are also somewhat subjective because the protein may adopt different conformations depending its relevant biological context. Thirdly, the solvent extractions may impose erroneous structure. $\endgroup$– user560Commented Oct 17, 2012 at 17:40
|
Show 1 more comment
4 Answers
$\begingroup$
$\endgroup$
0
6405 proteins mapping to 5220 genes, according to Ensembl.
In Ensembl's BioMart, you can select the PDB ID as external reference. Export the results and count the unique proteins/genes that have a PDB ID.
$\begingroup$
$\endgroup$
PDB is a good resource for answering such questions, since it will let you filter results by many additional parameters. To count and extract 3D structures of human proteins:
- Open
Advancedsearch tab of the PDB website. - Select
Biology->Source organismfrom the menu. - Type
Homo sapiens (human). - You can reduce redundancy by checking
Remove Similar Sequences at n% identitybelow. - Submit query.
To add further filters, click Refine Query with Advanced Search. There you can extract structures by deposition date, quality (eg. resolution or R-factors for structures solved by X-ray diffraction), ligands, enzyme classification, etc. (by checking Add Search Criteria)
Search for human proteins with removal of homologues with 90% identity cutoff fetches 7117 structures. The number of good quality X-ray protein structures (resolution < 2.5A) is currently 3964 (with the same identity cutoff).
You can then download the fetched list or create custom reports (menus below).
A good tool (also used by PDB) for generating non-redundant protein datasets is cd-hit.
answered Jan 15, 2012 at 0:06
$\begingroup$
$\endgroup$
From your comments it doesn't seem like you are adverse to writing some custom scripts so one option would be to take advantage of the NCBI Structure database. You can filter it by organism and then download the results as a text file / XML. If you need access to the raw PDB data you could then download the PDB archive and examine the ones in your filtered list.
$\begingroup$
$\endgroup$
PDBe's new search system is designed to answer just such questions http://www.ebi.ac.uk/pdbe/entry/search/index?organism_synonyms:HUMAN&view=macromolecules
shows there are 6964 unique human macromolecules with structure data in the PDB.
Of course, many will be fragments of proteins rather than the whole molecule.
