Making family trees from gene families

Since the time of Linnaeus, phylogenetic trees have been constructed using similarity of morphological characters to assess the relationships of organisms. The 1960s witnessed the evolution of an alternative (and now, familiar) system that uses the bases in nucleic acids or amino acids in proteins as the characters for tree-building. Today's genome projects provide vast amounts of sequence data, and might therefore be expected to give a solid foundation for this kind of molecular taxonomy. Of the organisms whose genomes have been completely sequenced, the vast majority are bacteria, and so one might suppose that bacterial taxonomy in particular would be more secure. Oddly enough, this is far from true. In this case, more data seem to mean more confusion, because different genes often define different trees. On page 108 of this issue, Berend Snel and colleagues tackle this problem by using a new approach based on gene families in completely sequenced genomes.