Have you tried just running them through BLAST? it would be the fastest way i can think of - just translate them into amino acid and use BLASTp to search for similarity to existing sequences that are already characterised and annotated. this should be very easy if the species you are studying is one of the well researched ones like mouse or human, and still easy otherwise, unless your study is on something really obscure.

website is very easy to use:

http://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE=Proteins

Hi Amr, thanks a lot for this suggestion. I already use the NCBI site. What I need is how to use the Gene Ontology to classify the sequences. Thanks

Perhaps the easiest would be to use BLAST2GO, this will give you GO and KEGG mappings.