Pdb Structures Go Terms Clustering And Distance

Question

[Reyhaneh]

Hi;

I have two questions regarding Go terms;

1) I would clusters of PDB proteins using their Go Terms. Are there any available datasets for this?

2) Are there any metrics that can give a distance between proteins based on their Go terms? (similar to the distance we can get between two protein sequences pairs). I need this metric to say how much two proteins are different functionally. thanks in advance.

Reyhaneh

Answer 1

[Khader Shameer]

1) You can derive GO terms for your PDB structures using multiple approaches and use such terms for clustering

• SCOP domain level annotation using SCOP2GO
  
• Transfer of GO terms to individual protein chains using PDB Advanced Search interface
  
• Use SIFTS annotations based on PDB chain to uniprot annotations (Use pdb_chain_go.lst)

2) If you have two GO terms you can compute semantic similarity (See this review for a background on basic concepts) between the terms. You can use packages like GoSemSim for such computations, extended list of tools here.

Answer 2

[Rph]

Hi Reyhaneh,

You can download a file of GO annotations (gene association file) for PDB structures from the UniProt-GOA ftp site here; ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/PDB/

The file format is described in the associated readme.

I hope this helps.

Rachael.