The review by B. Gaschen et al.(” Diversity considerations in HIV-1 vaccine selection,” 28 June, p. 2354) describes two computational methods (consensus and ancestral state) being considered for developing vaccine antigens against HIV. These methods attempt to minimize the amount of sequence divergence (distance) between the antigen and contemporaneously circulating viruses. Both methods do well at minimizing these distances when the sequences used to estimate the potential vaccine come from a symmetric phylogeny (panels A and B of the figure), similar to those examined in their Review. However, if sequences used to estimate the potential vaccine come from asymmetric phylogenies (panels C and D), then both methods generate sequences that poorly minimize these distances, making for a potentially poor vaccine antigen.

Four possible phylogenetic shapes and the resulting reconstructed sequences. The Ancestor (Anc) and the Center Of the Tree (COT) can only fall on an evolutionary path (ie, on a branch of the phylogeny), whereas the Consensus (Con) may not.