基于序列组成及联合信息论与机器学习方法对细菌必需非编码RNA的理论识别研究

Bacterial noncoding RNA is a newly discovered regulatory factor of gene expression in recent years, which plays crucial role in biological processes. Some ncRNAs are indispensable for the survival of living bacteria, which called essential ncRNA. Bacteria would not survive as long as the lack of any essential ncRNA. Thus, essential ncRNAs can be used as effective targets of antibacterial drugs and vaccine design. At present, the bacterial essential ncRNAs are identified by experimental methods. The experimental method is time-consuming and expensive. Furthermore, the experimental method is useless for some bacteria. This project aims at solving the defects of experiment method, and the problem of obtaining few sequence information from existing prediction method. This project develops the theoretical recognition model and software of bacterial essential ncRNA based on the adjacent bases, remote related bases and their implied relevant information. Hence, the information theory, the statistical discrimination method and machine learning are jointed to develop the model and software of predicting bacterial essential ncRNA. Using the software developed by this project, any bacterial entire essential ncRNA could be rapidly and accurately identified, which is almost without any cost. Hence, a bacterial essential ncRNA database will be constructed based on the predicting essential ncRNA. The smooth implementation of this project can provide efficient target of antimicrobial agents. At the same time, the theory research of essential ncRNA has important scientific significance for the theoretical research of the minimal genome.

细菌非编码RNA是近年来新发现的基因表达调控因子，在生物过程中扮演着重要角色。有一类ncRNA对细菌生长及其重要，称为必需ncRNA，细菌只要缺乏一个必需ncRNA就无法生存，因此必需ncRNA能作为抗菌药物及疫苗设计的有效靶标。目前，细菌必需ncRNA的识别采用的都是实验方法，实验方法需耗费大量的时间和经济代价，且对于某些细菌并不适用。本项目就是要针对实验方法的以上缺陷，和传统预测序列方法所得有效信息太少的问题，本项目拟基于相邻碱基、远程相关碱基及其隐含的相关信息等序列特征作为识别变量，联合信息论、机器学习和统计判别方法发展必需ncRNA理论识别模型和软件。利用本项目开发的理论识别软件可以快速、准确识别出任何一种细菌基因组所含有的全部必需ncRNA，经济省时，并建立必需ncRNA数据库。本项目的顺利实施可以为抗菌药物提供高效靶点，同时对最小基因组的理论研究有重要的科学意义。

细菌非编码RNA是近年来新发现的基因表达调控因子，在生物过程中扮演着重要角色。有一类ncRNA对细菌生长及其重要，称为必需ncRNA，细菌只要缺乏一个必需ncRNA就无法生存，因此必需ncRNA能作为抗菌药物及疫苗设计的有效靶标。本项目的主要研究目标是针对传统的理论识别必需 ncRNA 的方法所得的特征太少，拟联合信息论方法挖掘获取序列隐含深层信息发展出细菌必需 ncRNA自动识别软件和构建一个细菌必需 ncRNA 数据库。经过课题组几年的努力，该项目研究工作已经全部完成，下面将项目执行情况概述如下： （1）发展了基于序列特征的自训练的细菌必需 ncRNA 识别新方法，采用SVM模型进行分类器，跨种交叉验证的准确率在0.69 ~ 0.81之间。结果表明，我们所选择的特征对SVM方法预测必需ncRNA有较好的预测效果。发现细菌中必需的ncrna具有潜在的应用前景，可用于进一步的分析，如抗菌靶点的发现、最小基因组的构建和进化分析；（2）构建新的理论预测的必需 ncRNA 数据库，本研究通过收集已经测序的细菌基因组中包含的必需ncRNA，构建了必需ncRNA数据库。基于该数据库的数据，生物信息人员后续可以开发基因序列组成和序列衍生信息的必需ncRNA识别算法，同时可以利用其二级结构数据以提高相关算法的准确性；（3）进行其余重要的非编码DNA序列识别研究，包括生物合成基因簇识别软件（SynGC）开发，基于人工智能的准确快速预测Cas相关蛋白序列的算法，基因组重组热点位点序列的预测。利用本项目开发的理论识别软件可以快速、准确识别出任何一种细菌基因组所含有的全部必需ncRNA，经济省时，并建立必需ncRNA数据库。本项目的顺利实施可以为抗菌药物提供高效靶点，同时对最小基因组的理论研究有重要的科学意义。