• An exact match to their protein, i.e. you are trying to find your protein in the database, or perhaps a protein that contains your sequence.
  To do this you need to search your sequence against the database using BLASTP.
• You're looking for proteins that are quite similar to your protein, but not necessarily exactly the same.
  To do this, use PSI-BLAST. You can make the search more stringent (i.e. only find very similar sequences) by decreasing the e-value, e.g. 0.0001 means that the results have a 1/10000 probability of occuring by chance. If you increase the e-value, you will have a broader range of results, including less similar proteins, e.g. 10 means you would expect 10 sequences to be found by pure chance with the search.

http://cubic.bioc.columbia.edu/cgi-bin/var/database/pep/getPepEntry.cgi?orfId=q9npz4
http://cubic.bioc.columbia.edu/cgi-bin/var/database/pep/getPepEntry.cgi?fragId=human_fr57003
http://cubic.bioc.columbia.edu/cgi-bin/var/database/pep/getPepEntry.cgi?clusterId=14506

This CGI can be used to retrieve clusters (CLUP) or fragments (CHOP), using their NESG ID, e.g.:

http://cubic.bioc.columbia.edu/cgi-bin/var/database/pep/getPepEntry.cgi?clupNesgId=tt725
http://cubic.bioc.columbia.edu/cgi-bin/var/database/pep/getPepEntry.cgi?chopNesgId=tt725

You can also directly link to the SRS server by calling wgetz, the program which runs SRS on the web (this is what the above CGI does), e.g.:

http://cubic.bioc.columbia.edu/srs7bin/cgi-bin/wgetz?-newId+[chop-nesgid:tt725]

For detailed help on linking to wgetz, see the SRS on-line documentation.