Codon Substitution Models Based On The Residue Similarity And The Position Difference And Their Applications

Bioinformatics is one of active fields of multiple subjects, which involves mathematics, biology,computer science and other subjects. Phylogenetic analysis, as one of important research contents ofbioinformatics, makes the inference and evaluation of evolutional relationships from biologicalsequences by the probabilistic or statistical methods. We often use probabilistic substitution models todescribe the replacement between biological data units. Since1990s, the codon models combined withthe information of nucleotide replacement have received more and more attention, and have beenwidely applied.In this paper, the codon models considering the similarity of amino acids and the difference ofnucleotide positions in each codon are presented. In chapter2, considering the feasibility ofincorporating the amino acid information to codon models, we develop two codon substitution modelsbased on the similarity of amino acids. The first model considers the successive substitution on thesingle nucleotide position of single codon. Mutations involving more than one position in one codonare further considered in the second codon model. The continuous-time Markov process is used todescribe the substitutions among codons and the parameters of models are estimated by the maximumlikelihood method. Finally, the two new models are applied to three real data sets and compared withthe existing model to analyze the adaptability of models. The results suggest that the new codonmodels are more fit to the giving data comparing with existing codon models, and then present morereliable estimates of certain biologically important measures. In chapter3, according to the differenceof the substitution in nucleotide positions of each codon, we suggest a codon model that candistinguish these differences. We estimate the parameters of models by maximum likelihood methodand compare the new model with the old ones. In chapter4, to find the effect of gene conversion biasin detecting positive selection, we establish a substitution model based on the gene conversion biasfactor, and discuss how gene conversion bias affects the detection of evolution selection.