Zheng Wang¹, Francesc López-Giráldez¹ and Jeffrey P. Townsend^1,2

¹Department of Ecology and Evolutionary Biology and ²Program in Computational Biology and Bioinformatics, Yale University, New Haven, CT06520, USA.

Abstract

Achievement of the best balance between the accuracy and efficiency is always an important issue when searching a tree space of large data sets. In the 5th issue in 2009, Rodrigo et al used bootstrapped topologies as fixed genealogies to distribute an MCMC analysis across a cluster of computers, resulting in an efficiency yielding results 37 times faster than in the standard MCMC methods. Tree searches can seldom be guided with certainty, so that such snapshots sampling partial but “more-likely” tree space facilitated by parallel programs on computer clusters may provide great promise among a few choices that are computationally affordable when tree space is large and complicated.