Phylogenetic Tree Construction Based On Hierarchical Clustering

Molecular phylogenetic analysis is one of the most important fields inbioinformatics, which mainly studies homologous DNA or protein sequences and theevolution analysis for protein sequences or DNA sequences. So the phylogenetic treeis obtained which is reflected evolutionary relationship of protein sequences or DNAsequences. The research of phylogenetic tree reconstruction has a long time history; itplays a more and more important part in molecular evolution and phylogeneticresearch. The phylogenetic tree reflects the sequence of the species evolution, it helpsus to find the history and evolution mechanism of species. Phylogenetic treeconstruction methods can be obtained different categories of the different views, sowe can divide the phylogenetic tree construction methods into two categories: basedon mathematics clustering method and based on the optimal search method. As basedon optimal search methods are much slower than based on mathematics clusteingmethods, based on mathematics clustering method has wide applications.The hierarchical clustering method of based on mathematics clustering method isa generally method in phylogenetic trees construction, but the traditional mathematicsclustering method is built to base on sequence alignment. Therefore, in order to solvethis problem, we propose two new methods based on the hierarchical clustering, bothof which are without sequence alignment, and relatively intuitive, less calculation,easy to understand. These methods are the vertical and horizontal method and basedon changed weight by self-organizing map method.The vertical and horizontal method is also based on distance matrix method. It isbased on the distance matrix, and finds some points in these horizontal and someedges in these vertical to get a binary tree, and then we implemente clustering.Based on changed weight by self-organizing map method is proposed for basingon self-organizing map network. Self-organizing map network can be applied to bothclassification and clustering, and also be used in phylogenetic analysis andevolutionary tree construction. But when the self-organizing map network is usedphylogenetic analysis and evolutionary tree construction, the algorithm is complex, sowe propose this method which reduces the complexity of the algorithm. It is comparedwith based on distance matrix method, which should not be similar analysis toconstruct the distance matrix, and directly get a minimum weight tree. Finally weconstruct the phylogenetic tree by clustering. In order to assess the method feasibility of constructing the phylogenetic trees,we often use Neighbor.exe program of PHYLIP software to assess, we verify themethod feasibility by the experiment.