生物多序列相对特征分析方法的研究

多序列分析是生物序列分析的一个重要且具有广泛应用前景的课题。相对特征分析方法直接从原始数据出发，可以获得较多的信息，是解决序列比较问题的有效方法。然而，多序列比对模型，做为广泛使用的一种多序列相对特征分析模型，虽然存在着很难克服的缺点，但同时也缺少相应的其它可替代的多序列比较模型。多序列比对模型的不足主要是由于其相似性假设过于倾向序列的局部变化且无法考虑与序列的"序"相关的问题。因此，本项目主要研究多序列的非比对相对特征分析方法：提出具有数学意义、生物学意义的合理相似性假设，利用Burrows-Wheeler变换及序列间的公共子串来构建适合大序列、多序列的相似性对象并建立具有较低时间及空间复杂度的多序列比较模型，扩展应用于生物序列的相似性分析、分类与聚类分析以及物种进化等领域。

Analysis for multi-sequence is an important problem with the widely applied prospect in biology sequence analysis. Relative feature analysis method, starting from the original data directly is able to obtain much more information and an effective method for solving the sequence comparison problem. However, though alignment model of multi-sequence, which is widely used as an analysis model for relative feature of multi-sequence, has some weakness that is hard to overcome, there is a lack of other replacement for the multi-sequence comparison. The weakness mainly lies that its similarity hypothesis tends so much to the local mutations that its order information is neglected. Therefore, our project focuses on the analysis method of the alignment-free relative feature of the multi-sequence: raising the rational similarity hypothesis with mathematics as well as biology importance; setting up the similarity object fits for the large and multi-sequence and multi-sequence comparison model with lower time and space complexity by using Burrows-Wheeler transform and common words; broadening the fields applying similarity analysis, category and cluster and the evolution of the species of the biology sequence.