Mutual relationships of human endogenous retroviruses (HERVs) and their
similarities to other DNA elements are studied in this paper. We
demonstrate that a completely data-driven grouping is able to
reflect same kinds of relationships as more traditional biological
classifications and phylogenetic taxonomies. The clusters and their
visualization were computed with the Median Self-Organizing Map
algorithm of pairwise FASTA-based distances. The whole-sequence
distances are able to distinguish between the different known types
of endogenous elements, and exogenous retroviruses. The
HERVs become grouped meaningfully.