PhyloNet: Using Bayesian Inference and Network Theory to Map Life
In the highly influential Origin of Species, Darwin proposed the tree as the principle evolutionary model of life. An evolutionary tree is based on the vertical inheritance of genetic material. Over 1.5 centuries have passed since the publication of Origin of Species, and there are now many different methods to predict the likelihood of trees and categorize life on Earth. One newer method for predicting phylogenetic trees involves Bayesian inference. In addition to there being a variety of methods to create likely phylogenetic trees, there are also other models for the evolution of life that challenge the tree model. Luay Nakhleh, a computer science professor at Rice University, has created PhyloNet. PhyloNet is an open-source software package that more easily allows for the Bayesian analysis of phylogenetic trees, and also allows for modeling evolution through networks.
The Bayesian inference method for phylogenetic trees boils down to repeated applications of Bayes rule. The phylogenetic application of Bayes rule takes the following form:
Pr[Tree | Data] = (Pr[Data | Tree] * Pr[Tree]) / Pr[Data]
In other words, the probability of a tree given the data (posterior probability, or Pr[Tree|Data]) is based on the likelihood (Pr[Data|Tree]) and the prior probability (Pr[Tree]). To find the likelihood, DNA alignments are used to calculate how likely the data is according to the suggested tree – a better alignment leads to higher likelihood. The prior probability of a tree is the probability of some tree forming before making any observations. Most analyses set the prior probability of all possible trees equal. However, prior information about taxonomic relationships between the species being analyzed can allow scientists to set the prior probability of certain trees higher than others. Ultimately, the calculation of the posterior probability using Bayesian inference can allow researchers to arrive at phylogenetic estimates of the tree of life.
It is important to note that the tree model of evolution solely takes into account the passing of genes from parents to offspring. In reality, genetic information can be passed both vertically and horizontally. Organisms from simple bacteria to mice can hybridize with other organisms with whom they do not share a recent common ancestor. Accordingly, many scientists argue that networks – models in which a node may have more than one direct ancestor – more realistically model the tree of life. Using PhyloNet’s networks feature, bioinformaticians are beginning to represent the evolutionary history of life as a complex network rather than the vertical tree which has reigned supreme for over a century.
Source:
http://www.ecnmag.com/news/2014/11/program-models-more-detailed-evolutionary-networks-genetic-data