http://bioinformatics.oxfordjournals.org/content/26/21/2664.full?sid=1f67c073-33fd-40cc-8196-a8ea07ce3e9c

These results show an increasing proportion of newly determined sequences falling within existing islands, which may indicate an approach to the representative map of the protein universe. If this trend continues, by approximately 2017 at least 80% of new sequences will fall within an existing island (Fig. 4) that is, have a sequence identity >50% with sequences already present in the database.