Development of software for analyzing gene sequence data

点击翻译按钮获取中文摘要

We have developed maximum likelihood (ML) methods for inferring evolutionary trees from DNA and protein sequences. Since it is important to use realistic models for the substitution process in the ML analyzes, we have developed several models for the nucleotide and amino acid substitutions. Among them, a Markov model for amino acid substitution of mitochondrial DNA-encoded proteins might be the most important. In recent years, mitochondrial DNA sequences have been used widely in molecular phylogenetics. However, since the genetic code of mitochondria differs a little from the universal code, and since most of the mitochondrial proteins are membraneous, the transition matrix specific to the mtDNA-encoded proteins might be different from those of the models which are mainly based on the proteins encoded by nuclear DNA.Therefore, we estimated the matrix for the mtDNA-encoded proteins from 22 completely sequenced data of vertebrates by the ML.This matrix would be useful in analyzing mtDNA-encoded proteins.Since the ML method is computationally intensive, mainly due to a huge number of possible trees, there exists several limitations in applying the method to real biological problems. We have developed several methods to reduce the computational burden of the ML analyzes ; i.e., (1) approximate likelihood method to reduce the number of candidate trees, (2) star-decomposition method for topology search, (3) local rearrangement method for topology search.These results have been implemented in our program package MOLPHY,which are available from the network and have already been used widely in the world.