基于复杂网络的致病ncRNA预测模型及算法研究

More and more studies have shown that there is an important relationship between ncRNAs and diseases, so it is of great significance to study the forecasting models and algorithms of pathogenic ncRNAs. In recent years, with the continuous researches on complex network theory, the methods based on the complex network theory provide a new idea for further researches on prediction of pathogenic ncRNAs. The sparse characteristics of relationships between Known diseases and ncRNAs, between known diseases and diseases, as well as between known ncRNAs and ncRNAs, and the dynamical calculation problem of reconstruction of link-related weights in the construction of pathogenic ncRNA prediction model are the main factors to limit the performance of the prediction models and algorithms based on the complex Network. Therefore, based on the data resources of organic integration of heterogenous diseases and ncRNA databases, this project will firstly propose a dynamic calculation method of association weightes between diseases and diseases and between ncRNAs and ncRNAs, and then, a weighted complex network based on diseases and multiple types of ncRNAs will be constructed simultaneously. Thereafter, the pathogenic ncRNA prediction model will be established and trained based on the complex network theory. Finally, on the basis of the newly constructed prediction model and its key technologies, and a prototype system for predicting pathogenic ncRNAs based on the complex networks will be designed and developed. This project aims to lay a scientific foundation for human health and search for new molecules that can be possibly used for disease prevention, diagnosis and treatment.

研究表明ncRNA与疾病之间存在着重要关联关系，因此研究致病ncRNA预测模型及算法具有重要意义。近年来随着复杂网络理论研究的不断深入，基于复杂网络的方法为深入开展致病ncRNA预测研究提供了一个全新思路。已知疾病与ncRNA之间、疾病与疾病之间和ncRNA与ncRNA之间关联关系的稀疏特性，以及在致病ncRNA预测模型构建过程中关联权重的动态计算问题是制约基于复杂网络的致病ncRNA预测方法性能的主要因素，为此，本项目在有机整合异源疾病与ncRNA数据资源的基础上，首先研究一种疾病与疾病之间以及ncRNA与ncRNA之间的关联权重的动态计算方法，并在此基础上构建一个基于疾病与多种类型ncRNA的加权复杂网络。然后再基于复杂网络理论建立和训练致病ncRNA预测模型，并在此基础上设计并开发一套基于复杂网络的致病ncRNA预测原型系统。旨在为寻找可用于疾病诊断和治疗的潜在新型分子奠定科学基础。

近年来，越来越多的研究表明对致病ncRNA预测模型与方法展开系统深入的研究，不但可为疾病的预防和诊断提供潜在的新型途径，而且还可有助于新型靶向治疗方案的开发，具有重大的科学意义和广泛的应用前景。本项目系统研究基于复杂网络的致病ncRNA预测模型与算法，首先采用基于加权复杂网络的拓扑结构，研究基于复杂网络的致病ncRNA预测模型与算法设计中所必须解决的数据资源的收集与整合、疾病与ncRNA之间的关联权重的动态计算方法以及ncRNA与ncRNA之间的关联权重的动态计算方法等方面的新问题及关键技术难点；然后，基于复杂网络理论建立和训练致病ncRNA预测模型；最后，基于上述致病ncRNA预测模型及其关键技术，设计并开发出一套基于复杂网络的致病ncRNA预测原型系统。旨在为人类健康及寻找可用于疾病预防、诊断和治疗的潜在新型分子奠定科学基础。.基于以上研究内容，首先，在基于复杂网络的单种类型致病 ncRNA 预测模型的构建研究方面，先后提出了TCSRWRLD、CFNBC、LRWHLDA、PMFILDA、NBLDA、PCSLDA、BPNNHMDA、BWNMHMDA、NBLPIHMDA、BHCMDA、WINMDA、MSBMFHMDA、ICLRBBN等多种预测模型。然后，在基于复杂网络的多种类型致病 ncRNA 预测模型的构建研究方面，本项目先后提出了PADLMHOOI、LDLMD、FVTLDA、MCLCluster等多种预测模型。此外，还对microbes、drugs 和 diseases两两之间的潜在关联关系预测模型及数据库研究进展进行了全面综述。同时，还在前述理论研究的基础上构建一个用于MiRNA-disease潜在关联关系预测的数据库和可视化网络分析平台MDADP。.最后，基于上述理论创新，本项目先后培养青年教师或博士研究生5名、培养硕士研究生16名，发表高水平学术论文34篇，其中包括中科院一区TOP期刊论文3篇、JCR一区期刊论文15篇。此外，还获得授权国家发明专利4项。超额完成了本项目研究的预期目标。上述研究成果的取得，无疑将会为加速基于ncRNA生物标记或药物靶标的疾病的预防、准确诊断以及个性化治疗等方面的应用奠定科学基础。