微RNA与疾病关联关系的研究与实现

发布时间：2018-04-14 02:32

本文选题：生物信息学 + 微RNA　；参考：《中南大学》2014年硕士论文

【摘要】：在生物信息学领域,疾病与基因之间的关联关系是一个极其重要的研究方向。是研究人类遗传疾病的重要手段,掌握了复杂疾病的治病机理,就可以根据致病基因所对应的表型对症下药,或者避免遗传疾病的产生。随着越来越多的基因被发现,数据量的飞速提升,基因与疾病的对应关系将从生物学实验阶段步入新的计算机实验阶段。本文受到原有的微RNA-疾病对应关系算法研究的启发,提出了利用改进的随机游走算法在构建的疾病相似性网络上进行随机游走,对未知的微RNA-疾病关联关系进行打分排序,针对排序结果进行预测。主要的研究内容和成果如下：利用已发表文献挖掘经过生物学实验验证的微RNA-疾病的对应关系作为原始数据。利用基于Mesh概念的相似性算法计算相关疾病的相似性,用以构建疾病相似性网络。在传统的随机游走算法中默认初始节点游走至下一节点的概率是相等的,并没有结合网络本身的特性来进行判断,针对这一问题,本文提出了结合疾病网络的实际情况,构建基于节点重要性的疾病重要性矩阵,将节点重要性作为权重应用到转移概率矩阵中。同时,结合原始数据中的对应关系,本文为网络中已对应关系的疾病节点设置了一个评估参数,用来提升对应节点的转移概率,从而加大排序的正确性。实验结果验证了算法的可行性。实验结果得出本文365个微RNA关联的疾病排序,通过对每个RNA样本排名前20位疾病进行验证,证明了本文提出算法的准确性。同时与其他算法进行对比,验证了此算法的高效性,此外该算法的算法的AUC值达到了73.5%,相较于传统算法高出了一个百分点。算法的结果可作为生物学实验的重要参考依据进一步的应用于基因-疾病关联关系的实验之中。
[Abstract]:In the field of bioinformatics, the relationship between disease and gene is a very important research direction.It is an important means to study human genetic diseases. If we master the therapeutic mechanism of complex diseases, we can use the corresponding phenotypes of pathogenic genes to prescribe medicine, or to avoid the occurrence of genetic diseases.With the discovery of more and more genes and the rapid improvement of data volume, the corresponding relationship between gene and disease will move from the stage of biological experiment to a new stage of computer experiment.Inspired by the original algorithm of microRNA-disease correspondence, this paper proposes to use the improved random walk algorithm on the disease similarity network to rank the unknown micro-RNA-disease relationship.The results of sorting are predicted.The main research contents and results are as follows:The published literatures were used to mine the corresponding relationship between microRNAs and diseases verified by biological experiments as raw data.The similarity algorithm based on Mesh concept is used to calculate the similarity of related diseases to construct the disease similarity network.In the traditional random walk algorithm, the probability of the initial node walking to the next node is equal by default, and it is not judged by the characteristics of the network itself. In view of this problem, this paper proposes to combine the actual situation of the disease network.The disease importance matrix based on node importance is constructed, and the node importance is applied to the transfer probability matrix.At the same time, combined with the correspondence in the original data, this paper sets an evaluation parameter for the corresponding disease nodes in the network, which is used to enhance the transfer probability of the corresponding nodes, thus increasing the accuracy of sorting.The experimental results verify the feasibility of the algorithm.The experimental results show that 365 micro associated diseases are ranked in this paper, and the accuracy of the proposed algorithm is proved by verifying the top 20 diseases in each RNA sample.At the same time, compared with other algorithms, the efficiency of the algorithm is verified. In addition, the AUC value of the algorithm reaches 73.5%, which is one percentage point higher than the traditional algorithm.The results of the algorithm can be used as an important reference for biological experiments.
【学位授予单位】：中南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R3416;TP301.6

【参考文献】