Inference of molecular phylogenetic tree based on minimum complexity principle

在从分子数据重建系统发育树中，人们已经指出，用传统的最大似然法等方法很难正确地重建多分支的系统发育树。为了解决这一问题，我们一直致力于开发一种新的基于最小复杂性原则的系统发育树重建方法，该原则在归纳推理中得到了广泛应用。在本研究中，我们定义了一个新的复杂性概念，我们称之为“基于经验模型的复杂性”，并提出了用于重建分子系统树的“基于模型的最小复杂性”准则。该方法通过与树的拓扑结构、分支长度和模型与似然函数测量的数据之间的适合度三个项来描述分子系统发育树的复杂性，并通过计算机模拟与最大似然法和Akaike信息法(AIC)比较，考察了该方法在有根和无根系统发育树估计中的有效性。结果表明，与传统的最大似然法及其改进方法相比，最大似然法具有更好的渐近性，AIC方法在多叉树作为树的候选拓扑和/或长DNA序列可用于重建系统发育树(超过3000个碱基)的情况下，避免了树模型相对于现有物种DNA序列信息量的过度复杂性。因此，它可普遍用于构建具有任意多分支的系统发育树。

In reconstruction of phylogenetic trees from molecular data, it has been pointed oat that multifurcate phylogenetic trees are difficult to be correctly reconstructed by the conventional methods like maximum likelihood method. In order to resolve this problem, we have been engaged in developing a new phylogenetic tree reconstruction method based on the minimum complexity principle widely used in the inductive inference. In this study, we defined a new concept of complexity, which we call "empirical model-based complexity" and proposed "minimum model-based complexity(MBC)" criterion for reconstructing molecular phylogenetic tree. This method describes the complexity of molecular phylogenetic tree by three terms which are related to the tree topology, the branch length and fitness between the model and data measured by likelihood function.The Computer simulation is used to investigate the efficiency of this method in estimating rooted and unrooted phylogenetic tree in comparison with those of maximum likelihood method and Akaike information method(AIC). The results suggest that the MBC method has a good asymptotic property compared with traditional maximum likelihood method or its modification, AIC method in the case that the multifurcate tree is considered as a candidate topology of the tree and/or long DNA sequences could be used in reconstructing phylogenetic tree (over 3000-bp), because it avoids excess-complexity of the tree model in relation to the amount of the information available from DNA sequences of current species. Therefore it could be generally used for reconstruction of phylogenetic tree having arbitrary multifurcations.