Hunting For A Protein Based On Its Domain Architecture

Question

[jvijai]

Hi,
(Disclaimer: I am a novice in protein structure informatics)

I am trying to find proteins with all of the following features in any annotated protein database. It has:
1. One ankryin repeat region,
2. One leucine-rich repeat (LRR) domain,
3. One kinase domain,
4. One DFG-like motif , 5. One RAS domain,
6. One GTPase domain,
7. One MLK-like domain,
and 8. a WD40 domain

Ankryin-LRR-Kinase-DFGlike-RAS-GTPase-MLKlike-WD40

Where do I start?
What are the logical steps to zero in one possible candidates?

Thanks
:)

[jvijai]

Thanks everyone. Will try out these methods.

Answer 1

[Lars Juhl Jensen]

You can use the Architecture analysis functionality of SMART for posing this kind of questions. However, it does not know of any proteins that match your long list of requirements.

Unrelated to that, requirements 5 and 6 would seem redundant, since the RAS domain is a GTPase domain.

Answer 2

[Michael Schubert]

I am not aware of any tool that does this in one step. Maybe there is one that I'm not aware of (CLCbio?) and other users can provide input for.

Otherwise, I would say your best bet is to:

1. Find models of the domains you are searching for. Depending on whether you would like to use RPSBLAST or HMMER, these might be in the NCBI conserved domain database (likely to be more complete) or e.g. in the PFAM HMM library
2. Find and set up a sequence candidate database to use for model testing. This might be, creating a local BLAST db out of SwissProt sequences (your computer is likely not able to handle NCBI's nr database)
3. Search your candidates using each domain model with the appropriate thresholds
4. A candidate that is in all results (setting aside how likely this is for your domains) will have all domains

Note that some steps in there will require the knowledge of how to use the command-line BLAST or HMMER tools. Shell-level scripting will be a big advantage. This may or may not be a problem for you.

Answer 3

[Fred Fleche]

How To Retrieve Human Proteins Sequence Containing A Given Domain

And I was quite happy with Marina's method using the advanced search feature at Uniprot.

http://www.uniprot.org/uniprot/?query=domain%3AANK+AND+domain%3ALRR&sort=score Feel free to add more domains.