节点文献

功能基因组学和化学信息学协同的药物研发数据挖掘方法

Synergistic Data Mining Methods between Functional Genomics and Chemoinformatics in Drug Development

【作者】 俞书浩

【导师】 李亦学;

【作者基本信息】 上海交通大学 , 生物学, 2013, 博士

【摘要】 药物研发是一个需要涉及到数学、化学、生物学、计算机科学等各个专业领域的综合过程。新药研发中的关键一环是找到药物的作用靶点和作用机理,人类基因组草图绘制完成后,人们普遍认为,利用大规模高通量的组学技术和生物信息学技术,药物机理的发现会呈现一个几何级的增长。但事实上药物研发的速度这些年来并没发生太大的改善,究其原因,主要在于对高通量数据分析方法的滞后。传统的分析方法只是在药物实验组的基因表达数据和对照组表达数据比较,选出其中表达量水平差别比较大的差异基因,然后在里面挑选基因来一个个验证,这种做法无异于大海捞针。为了解决这个问题,本论文首先设计了一种大规模的跨物种跨平台整合药物转录组数据来挖掘药物机理、副作用或老药新用以及分析组合用药的方法CGEMining,通过将各种不同实验中的动物疾病模型的转录组数据和药物干扰的转录组数据关联起来,并且关联到细节的功能模块,然后从这些关联中挖掘出已知功能的小分子的未知的能治疗其他疾病的功能,以及这些小分子的一些未知的副作用,并且深入探讨其中的机理。其中使用到的主要方法是直系同源转换、GO富集分析以及基于KS检验的比较方法。而对于一些未知机理的药物小分子,也可以利用CGEMining方法通过比较分析,挖掘出其潜在的作用靶点,然后利用化学信息学中的分子对接方法来进一步确认CGEMining挖掘的结果,本文中使用了一组针对糖尿病的新的药物分子作为例子来说明此过程,并且发现PPARgamma是其潜在的作用靶点。同样对包含多种成分的中药复方,CGEMining方法也能挖掘出其复杂的作用机理,而且能够发现其新的功能。本文中挖掘出了治疗肝纤维化的复方扶正化瘀胶囊的降血糖等的新作用。除了CGEMining方法之外,本论文中还开发一种新的基于蛋白序列数据和化学信息学协同的药物数据挖掘的方法VIScreen,通过利用化学信息学的结构信息来建立病毒蛋白序列和药物小分子之间的联系,以达到快速挖掘病毒药物与靶点结合的信息以及辅助新药研发的效果。最后,本论文还介绍了一种单独使用基因转录组数据的药物小分子机理挖掘方法,是以转录因子为核心进行差异共表达分析,构建共表达的网络模块,通过比较不同网络模块之间的差异来分析两种药物在网络药理学上的不同。

【Abstract】 Drug development is an integrated process involving mathematics,chemistry, biology, computer science and so on. The key points in thedrug development process is to find drug targets and the mechanism ofaction. After the completion of the Human Genome Project, peoplegenerally believed that along with the large-scale high-throughputgenomics and bio-informatics technology, the discovery of new drugtarget will presents a geometric growth. But in fact, the speed of drugdevelopment all these years did not have much improvement. Thereason mainly lies in the lag of high-throughput data analysis means.The traditional method for the analysis of gene expression data is toseek drug targets within a lot of difference expressed genes. Thispractice is tantamount to fish for a needle in the ocean.To solve this problem, a large-scale cross-species data miningmethod, CGEMining, is developed in this paper to dig information ofdrug mechanism, drug repositioning, side effects, and combination ofmedication. By combining various transcriptome data of disease ordrug experiments on animals, CGEMining could dig out drugrepositioning, side effects and drug mechanisms from the drug-diseaseor drug-drug relationships. For some small molecule drugs withunknown mechanism, CGEMining methods could help find thepotential targets by comparative analysis, and then docking will be used to confirm the results. This process is illustrated by a case ofanalyzing the target of diabetes drugs. For some herbal compound thatcontains a variety of ingredients, CGEMining method could not onlydig out its complex mechanism, but also discover its new features. Inaddition to the CGEMining method, another method, VIScreen, willalso be introduced in this paper, which combines viral protein sequencedata and chemical informatics. VIScreen takes chemoinformaticsstructure information as an intermediary for creating linkages betweenviral protein sequences and drug small molecules and then mine thepotential combination of viral drugs and protein targets. Finally, thispaper also introduced a causal co-expression network analysis methodto study the difference between two drugs in network pharmacology.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络