基于蒙特卡洛理论的基因序列分析与仿真

发布时间：2018-03-13 15:06

本文选题：系统发育　切入点：序列分析　出处：《广东工业大学》2016年硕士论文　论文类型：学位论文

【摘要】：生物大数据背景下的数据规模对传统的生命科学分析和实验方法提出了严峻的挑战,各生物数据项之间复杂的关系也是传统的研究模型难以描述和处理的。为了解决这个问题而兴起的生物信息学正是着重借助计算机的强大性能,通过针对问题构建各种模型并参考其他领域经典的算法,为生命科学的研究和发展提供帮助和支持。为了推动生命科学的发展,对传统方法的继承和改进十分必要的。考虑到传统方法没有在计算中表征序列的具体功能,在比较大的规模的数据集中传统方法的分析结果与实际观测结果的误差较大,因此需要在计算中考虑序列具体功能含义。而生物数据分析本身是非确定性的问题,使用确定性分析的计算方法求解在理论和实际中都存在不同程度的误差,考虑使用非确定性分析的模拟归纳方法有效的提高结果的准确度,并降低冗余计算消耗的时间。针对这两个问题,本文主要进行了两项工作。首先是借鉴了数据挖掘技术在文本分析和自然语言处理领域的分析方法,通过对数据集的预处理建立向量库以分析保守序列的结构模式,从而确定序列之间置信度概率；下一步工作则是针对这个概率,使用蒙特卡洛方法进行发育分析模拟,并在大量的模拟实验中确定最优解。针对传统的系统发育分析方法中没有考虑到具体序列功能和含义的问题,本文在进行系统发育分析的工作中引入了携带具体序列功能的保守片段的研究方法,不但可以在分析中计算具体序列的含义,还能将长度在百万至千万位的序列分析降低到长度仅在数十至上百的模块分析,在降低计算时间和空间需求的同时,提高分析结果的置信度和可靠性。实验结果表明,本文提出的方法的时间复杂度和空间复杂度均优于传统的方法,而且相比于传统方法本文增加了对具体序列功能和含义的考虑和分析,得到的结果不但符合数据的校验结果,更可以提供具体序列功能性的解释,在置信度和可靠性上都超过传统的方法。使用蒙特卡洛模拟分析更符合自然进化的模式,得到的结果更完整、全面,将会对系统发育和生命科学研究和建设提供更可靠的分析工具,以弥补传统的分析方法的不足之处。
[Abstract]:The data scale under the background of biological big data poses a severe challenge to the traditional life science analysis and experimental methods. The complex relationships between biological data items are also difficult to describe and deal with in traditional research models. Bioinformatics, which has arisen in order to solve this problem, focuses on the powerful performance of computers. In order to promote the development of life science, we build various models and refer to classical algorithms in other fields to help and support the research and development of life science. It is necessary to inherit and improve the traditional method. Considering that the traditional method does not represent the specific function of the sequence in the calculation, the error between the analysis result of the traditional method and the actual observation result is large in the larger data set. Therefore, it is necessary to consider the specific functional meaning of sequence in calculation, and biological data analysis itself is a non-deterministic problem, and the calculation method of deterministic analysis has different degrees of errors in both theory and practice. Considering the use of non-deterministic simulation inductive methods to improve the accuracy of the results and reduce the time of redundant computation. In this paper, two main works are carried out. Firstly, we use the analysis methods of data mining technology in the field of text analysis and natural language processing for reference, and build a vector library to analyze the structural patterns of conservative sequences by preprocessing the data sets. In order to determine the probability of confidence between sequences, the next step is to use the Monte Carlo method to simulate the development of the sequence. The optimal solution is determined in a large number of simulation experiments. The traditional phylogenetic analysis methods do not take into account the function and meaning of specific sequences. In the work of phylogenetic analysis, this paper introduces the research method of conservative fragments with specific sequence function, which can not only calculate the meaning of specific sequence in the analysis. It can also reduce the length of sequence analysis from millions to tens of millions of bits to only a few tens to hundreds of modules in length, while reducing the need for computing time and space, while improving the confidence and reliability of the analysis results. The experimental results show that, The time complexity and space complexity of the proposed method are better than that of the traditional method, and compared with the traditional method, the function and meaning of the specific sequence are considered and analyzed in this paper. The results not only accord with the results of data verification. It can also provide a functional explanation of specific sequences, outperforming traditional methods in terms of confidence and reliability. Using Monte Carlo simulation analysis is more consistent with the natural evolutionary model, and the results are more complete and comprehensive. It will provide more reliable analytical tools for phylogenetic and life science research and construction to make up for the shortcomings of traditional analytical methods.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：Q811.4;TP311.13

【相似文献】