Classify Proteins Into Classes

Question

[miquelduranfrigola]

Hi all,

I am aware that a How To Functionally Classify Proteins? was asked a few years ago in this forum, but I did not find a clear answer.

I have a set of ~1,000 proteins (UniProt ACs) and I would like to group them into a few (10-20) families or classes. I know that servers like InterPro are useful, but their definition of "family" is too specific for my interests. Rather, I'd like to classify my proteins into something like "Membrane receptor", "Ion channel", "Structural", "Kinase", etc.

Are you aware of any resource that provides this kind of annotation? Would the first level of GO Molecular Function do?

Thanks in advance,

Miquel

[David Westergaard]

You do realise some protein will fall into multiple categories if you pick such a simple scheme, right? An example of this would be EGFR, a membrane-bound kinase receptor.

Also, I'm not really sure if you would get what you want from the first level of GO molecular functions. As another member pointed it, it is extremely general. Why would you want to classify them anyway? If you add this information it might be easier to answer your question.

[PoGibas]

As you asked for example, I have similar classification problem (question) - how can classify (extract) all the genes (proteins) that have zinc finger domain?

[David Westergaard]

Search The Protein Domain By Sequence In Batch Work

Answer 1

[Alex Paciorkowski]

Would the first level of GO Molecular Function do?

Depends on what you need this annotation for. If InterPro's classification are too specific, that first level of GO Molecular Function can be...very general. So it depends on what you will be using your annotation for. Usually it is advisable to use an existing ontology for any kind of annotation...but rules were made to be broken.