Application Of Improved Phylogenetic Profile Method In Protein Functional Prediction

With the arrival of post-genome era, the focus of bioinformatics increasingly has been transferred from the genome sequencing to genome functional annotation. The traditional method that based on homology can not meet people's needs in the accuracy because of their own shortcomings. While non-homologous methods gradually become more and more important. Being differ from homologous methods, non-homologous methods predict the function of genes or proteins according to their evolution relevance. Among so many non-homologous methods, phylogenetic profile method is the most widely applied one for its great research value.Phylogenetic profile method aroused people's attention as soon as it was proposed by Pellegrini in 1999, and many researchers did a lot of work to improve on it in the next few years. The improvements on phylogenetic profile method implemented mainly on the three steps including the selection of the reference genome, the foundation of phylogenetic profile and the analysis of the profiles'similarity. The phylogenetic profile method still has many deficiencies, for example, there is a lack of certain standard on the selection of reference genome, the classical clustering algorithm has not yet been used in the clustering of similar profiles effectively. To these deficiencies, two improvements on phylogenetic profile method were proposed in this paper. First, constructing weight based phylogenetic profile, which decreased the excessive dependence on the selection of reference genome. Second, joining the hierarchical clustering algorithm and K-mean clustering algorithm which are the most popular classical clustering algorithm to obtain better cluster result.