Abstract
Nowadays, new technique reduces the cost and time of sequencing projects. There are around four thousand complete genomes and fifteen thousand ongoing sequencing projects in the world. Dealing with this huge data becomes an interesting topic. This dissertation focuses on three topics including the usage of complete genomes, incomplete genomes and comparisons of gene interactions. On the complete genome analysis, information can be retrieved directly from genome analysis instead of functional and metabolic experiments. For example, the possible metabolic pathways of an organism can be identified by homologous search from enzymatic database. However, the genomic analytic methods have limitations, such as the computational time causes and predicting accuracies. For this reason, we developed a strategy called probe-set method extracted information from whole genomes. It has high sequence-error tolerances and it also costs little computation time to analyze hundreds of bacterial genomes. Classification tree rebuilt based on the result reflects the metabolic characteristics. Furthermore, to evaluate our classification result, we developed a simple tree comparing method based on topological similarities of queried trees. This method provides tree comparing standard for further tree reconstruction methods. For incomplete genomes, except for the whole genome analysis, the specific regions within the chromosomes called genomic islands bring external functionalities into bacteria, such as antibiotic resistant capabilities and pathogenic abilities. Discovering of genomic island (GI) sequences is important in most bacterial genome projects. We built a Web server that offers GI prediction for incomplete genomic sequences. It provides pre-analytic information that assist researchers to finish their genome projects. For more specific analysis, such as systematic comparison of signaling transduction mechanisms, we developed the comprehensive gene-gene interaction network database and user-friendly graphic analytic platform, called PCmap. To sum up, we developed methods from global genomic point of view into gene cluster and gene interaction analysis. We believe these methods and Web servers can be useful for understanding and discovering the metabolic and functional relationships of organisms.