you can do with swissprot clustalW

you can go though www.expasy.org and find so many tools for your queiry

Align the sequence using MAFFT.. you will get some clue about the common motifs.

Its lengthy process but useful for you first submit an individual sequence in CATH and find the domain information

This is really good too:

http://www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml

Are you looking to do an alignment of your sequences? This will basically show you if you have any areas that are conserved across all 7 proteins.

You will need a fasta fill that contains each protein, like this:

>Protein 1

some sequence

>Protein 2

a different sequence

You can then using software like muscle or CLUSTALW to align your proteins. You can then view your alignment in UGENE or Seaview or some other view of your choice.

Thank you all for your reply...

More about CDD (see above.)

The National Center for Biotechnology Information (NCBI) has a searchable Conserved Domain Database. Submit AA sequence(s) in FASTA format at

http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi

then compare what domains the CD-search finds in your seven.

To make best use of it, search both with & without Low-Complexity Filter to catch borderline or weak domain instances. Run with different Expect Value threshold values for the same reason.

This kind of search is very fast, so run with "Force live search" turned ON to be sure to get everything.

I got a lot of insight into bHLH (basic helix-loop-helix) proteins I was studying with this tool. The Low-Complexity Filter on/off parameter was particularly useful because the "basic" portion, especially if very strongly basic, could be mistaken for being a low-complexity (and thus not interesting) area.

The place to start is InterPro. It contains e.g. Pfam and SUPERFAMILY, Panther, etc.

I agree with many answers above, but maybe to summarize:

1. I would make an alignment with any tool (there're plenty) -> the common motif is the candidate for being the domain you're hunting for

(1a. you can also predict the motif using MEME (http://meme.nbcr.net/meme/). Technically, it is easier).

2. I'd take the found common motif and treat it in any of the ways:

2a. Go to Prosite. There you can make a consensus of your motif and see if this consensus is already known. Otherwiese, just run a prediction directly, there's an option for that.

2b. Go to InterProScan, as suggested above. It already contains PFAM and other domain databases, so no need to scan them in addition.

2c. I would maybe also run the prediction in NCBI, as suggested by Michael Collins.

You can also do the whole job the other way round: start with the InterProScan prediction of known motifs in each of your proteins (you have only 7, so not a big deal). Then compare the results: if any domain is shared, you will see that.

If your sequences contain an unknown domain, you can only predict the common motif (with aligning or MEME), but then you will need to run some wetlab experiments to see if your assumptions about its possible function are correct.