Brian Foley, Grian Gaschen, Vadim Zalunin, Bette Korber and the HIV database staff.

The talk will address three main topics. The first is the global and subtype variation represented in the database. GenBank now requires authors add a country of origin to entries, and we have been trying to track this as well for many years now, so now we have country code information for 27,000 sequences and subtype information for 25,000 sequences. The database has a new tool for geographic displays of this information on global or continent maps. While this tool is useful for assessing variability from a region and our extent of knowledge about a region, it is not appropriate for epidemiological summaries because of strong sampling biases. The second topic is global evolution of the V3 loop. The V3 region of gp120 is the still the most sequenced region of the genome, and half the available sequences span the V3 loop. The distinctive behaviors of V3 evolution in the C subtype and D subtype relative to all other subtypes is strongly apparent: C subtype sequences are much more highly conserved within the V3 region, and D subtype radically divergent, with many more positively charged V3 loops. This pattern is now holding for over 400 sequences from C-subtype infections and close to 200 from D-subtype infections, and is specific for the V3 loop – in contrast to the V3 loop, regions proximal to the loop are evolving in a subtype-independent manner. The third topic is subtypes and known epitope variation. The point is made that while there may be many cross-clade conserved epitopes, there are many others which have diverged in a subtype-specific way. The proteins with the greatest level of conservation, p24 Gag and pol-encoded proteins, may potentially serve as excellent antigens to stimulate broad CTL responses, but the breadth of the response to more variable proteins like gp120 Env and Nef have many potential eptiopes lost to the extreme variation. The potential of a consensus sequence or ancestral sequence to allow for stimulation of a greater percentage of potential targets was proposed, and the relative conservation of known eptiopes was tallied. A consensus protein, the consensus of the consensus of each clade, proved to include many more conserved epitopes than did cross-clade reagents, but fewer than within clade reagents, with a most pronounced potential benefit in Env and Nef.