protein classification based on functional regions

Question

[asdfxzsdf2]

Is anything out there in terms of classifying proteins based on functional regions?

For example, suppose you have 2 very similar sequences A and B, perhaps in the same protein family, and an unknown sequence C. To determine which of A and B the unknown protein C is, you could look beyond the overall alignment to the functional regions.

Answer 1

[Jean-Karim Heriche]

PFAM, InterPro and CATH

[asdfxzsdf2]

Hi Jean-Karim,

It looks like PFAM and Interpro both classify proteins into families making use of functional regions, without distinguishing between proteins within the same family. CATH displays families as well as the found domains.

I'm looking for a greater level of specificity achieved by using protein attributes. For example, uniprot has a 'sites' section which lists feature keys (like active sites, element binding sites, etc) in addition to their sequence positions.

Can you not differentiate between proteins of the same family by looking at conservation across these sequence positions? I'm looking for a tool which has done this.

I'm a student looking for a way to contribute in the area of annotation, your opinion is greatly appreciated.

Thanks

[Jean-Karim Heriche]

I am not sure I get which level of specificity you need. If UniProt has the info you need, why not use it ? UniProt integrates information from Interpro and PFAM among others. If you want to distinguish between members of a protein family such as differentiating between paralogs, you may need to do a phylogenetic analysis.