Abstract. In this paper we describe a method for computing genomic midpoints, which minimizes the sum of the distances to three given genomes, and apply the method to three data sets which have been analyzed previously by other methods: a human-lemur-tree shrew comparison that estimates the ancestral primate karyotype, several versions at different resolution of a human-cow-cat comparison, and a human-mouse-rat map with 424 markers. Rather than report one particular midpoint as the "right answer," as previous analyses have done, we produce a large number of possible midpoints. As in the reconstruction of phylogenies, comparing these solutions helps us understand the confidence we can have in various features of our predicted midpoint, and in our estimates of the number of events that occurred on various lineages.
Preprint
Supplementary Materials: Description of the algorithm
Comparative map
Computer code is available from
Yannet's Cornell Web Page