Detecting Character Dependencies in Stochastic Models of Evolution

摘要

Abstract Stochastic models of biological evolution generally assume that different characters (runs of the stochastic process) are independent and identically distributed. In this article we determine the asymptotic complexity of detecting dependence for some fairly general models of evolution, but simple models of dependence. A key difference from much of the previous work is that our algorithms work without knowledge of the tree topology. Specifically, we consider various stochastic models of evolution ranging from the common ones used by biologists (such as Cavender-Farris-Neyman and Jukes-Cantor models) to very general ones where evolution of different characters can be governed by different transition matrices on each edge of the evolutionary tree (phylogeny). We also consider several models of dependence between two characters. In the most specific model, on each edge of the phylogeny the joint distribution of the dependent characters undergoes a perturbation of a fixed magnitude, in a fixed directi..." /> rel="meta" type="application/atom+xml" href="http://dx.doi.org/10.1089%2Fcmb.2015.0099" /> rel="meta" type="application/rdf+json" href="http://dx.doi.org/10.1089%2Fcmb.2015.0099" /> rel="meta" type="application/unixref+xml" href="http://dx.doi.org/10.1089%2Fcmb.2015.0099" /> 展开▼