+ See all authors and affiliations

Science  16 Aug 2002:
Vol. 297, Issue 5584, pp. 1097b
DOI: 10.1126/science.297.5584.1097b

The history of life on earth and the relationships of organisms to one another are commonly expressed in the form of branching phylogenetic trees. However, finding the optimal phylogenetic tree for a group of organisms is a computational headache that becomes ever more painful with the number of taxa involved. The number of possible trees runs into billions with as few as 10 taxa, so phylogeneticists have developed “maximum likelihood” methods that find one or several near-optimal trees much faster than the perfect, correct tree. Even so, such methods are expensive in terms of computational time and power.

Lemmon and Milinkovitch have developed an algorithm that promises radical improvements in the speed and efficiency with which maximum-likelihood trees are found. Their method, called the metapopulation genetic algorithm, yields maximum-likelihood trees from nucleotide sequence data from hundreds of taxa on a normal desktop computer in a working day. If widely applicable, this method opens opportunities for a surge of studies of large phylogenies and a new depth of understanding of the intricacies of the evolutionary relationships between organisms. — AMS

Proc. Natl. Acad. Sci. U.S.A. 99, 10516 (2002).

Related Content

Navigate This Article