【作者】 张琦；

【导师】 邹北骥；

【作者基本信息】 中南大学 ， 计算机应用技术， 2009， 硕士

【摘要】 蛋白质三维结构与功能的关系问题是当前生命科学领域一个重大科学问题,蛋白质三维结构相似性比较则是探索蛋白质结构与功能关系的一种重要手段。基于计算机图形学方法利用分子结构上的相似性来进行分子性能上的比较是目前分子相似性比较的一种重要方法。但已有的表达和描述比较模糊,许多方法都可以得到一种相似性比较的结果,然而其计算复杂度比较复杂。使用计算机图形学辅助其他学科找寻相似性具有很多优势,其中最大的优势是图形学有强大的空间表达能力。本文提出了一种基于计算机图形学的蛋白质结构相似性比较方法,通过使用空间球极坐标三个分量划分蛋白质所在区域,得到蛋白质基本组成元素的空间密度特征,进而判定2个蛋白质结构的相似性。同时,采用空间球极坐标下主成份分析法(principle component analysis,PCA),在进行相似比较前将2个蛋白质大分子沿主轴对准,保证了相似度计算的几何不变性。基于本文方法设计了一个蛋白质分类实验,结果表明,该方法可作为各种结构蛋白质相似性比较的有效辅助手段,不仅能较好地反映蛋白质分子的空间结构特征,而且还有令人满意的时间复杂度,对大分子的相似性比较及进一步分类将有重要意义。更多还原

【Abstract】 The intrinsic relationship between the function of a protein and its structure is an important issue in the study of contemporary life science and the similarity comparison of protein structures can provide effective hints in such study. The similarity comparison of molecular structures based on computer graphics provides a method of evaluating similarity of molecular via its performance evaluation. The expression and description of the similarity of molecular structure are still very vague today. There are many methods bringing about a result of similarity comparisons. However, their computation complexities are complicated.It is advantageous for the discovery of similarity by computer graphics which is used to aid other disciplines, and the most advantages is that graphics has a great ability of space expression. A novel algorithm for comparing protein structure similarity is proposed. By using spherical polar coordinates to divide the space of proteins, the space density of main elements in protein molecules could be obtained easily and quickly. With the help of density fingerprints, the similarity between the proteins could be computed. At the same time, this essay has exploited PCA in space spherical Polar coordinates. It makes principal axes of 2 proteins macromolecular in alignment before the evaluation of similarity in order to guarantee the geometrical invariability of similarity computation.This essay designs an experiment about the classification of proteins. The experimental results show that the proposed method could be used in similarity determination for proteins. Moreover, the results of classification of the proteins of different functions based on their density fingerprints seem to be quite reasonable. Thus, the approach developed in this thesis might be of significance in similarity determination and further classification for macromolecules.更多还原