中国科学院昆明动物研究所机构知识库

转录组是在某一特定时间和空间下组织或细胞内基因所有转录产物RNA的总和。转录是基因表达的第一步，也是关键的调控环节，通过基因表达可以将基因携带的遗传信息转变成可识别的表型特征。转录组学是从RNA水平上研究细胞内的基因表达情况，是探究细胞表型和生物分子功能的有效手段。随着二代测序技术发展，各种人类疾病或者生理状态下的转录组数据不断积累，利用不同的数据挖掘方法从基因表达水平探究正常的生物过程和疾病的病理机制成为了研究的热点和难点。为了更好地解决复杂的生物学问题，需要对海量的转录组数据进行整合分析，通过合理有效的方法或者算法，合并多个针对某一疾病或生理状态的独立研究数据集，从而扩大样本量，提高整个分析结果的可信度和一致性。因而，本研究围绕着转录组数据，通过整合其他不同种类的数据和不同的数据挖掘方法，探究老年人群常见的两种复杂疾病——阿茨海默症和中风，进行了以下3个相对独立的研究，并取得了一定的进展。 1. 衰老相关基因NRXN3的低表达增加阿茨海默症的发病风险阿茨海默症(Alzheimer’s disease, AD)是一种常见的神经退行性疾病，衰老被认为是AD最大的一个风险因素，年龄越大的人群发病率越高。本研究的目的是探讨衰老是如何导致AD风险增加的，找到正常衰老人群和AD病人中常见的协同并发的变化，找到具有衰老依赖性的并促进AD发生和发展的基因。我们分别搜集了3组AD和1组衰老相关的表达数据集，并将衰老数据集分青年组、中年组和老年组。通过不同的算法对两类表达数据集进行分析，MetaMA方法计算AD基因的差异表达，SAM方法探寻衰老数据集中与年龄（衰老）密切相关的基因，包括与年龄正相关的和负相关的基因。本研究发现在AD病人的差异表达基因中，有1291个基因是与年龄有密切相关性的，基本上在AD病人中表达上调的基因与衰老具有正相关性，同时在AD病人中表达下调的基因与衰老具有负相关性，即这些基因在AD病人和健康老年人群中表达趋势基本一致。对这1291个基因进行KEGG通路的富集分析，大部分的基因富集到了神经退行性疾病、细胞凋亡、阿茨海默症、泛素介导的蛋白水解等KEGG通路中，说明这些基因在正常衰老个体中增加了老年人人群AD的患病风险。整合与AD发病相关的基因单核苷酸多态性，发现在1291个共有失调基因中有32个基因携带有与AD发病相关的单核苷酸多态性。其中基因NRXN3是AD中排在第二位的差异表达基因（下调），同时NRXN3的表达与年龄呈负相关。因而，我们着重探讨了NRXN3及其相关基因的功能网络，发现NRXN3在突触功能和神经活性方面发挥重要的作用，在健康衰老人群中表达下调的NRXN3是促使AD发生重要的遗传风险因素。2. 通过共表达网络探寻阿茨海默症中的潜在致病基因阿茨海默症（AD）是一种进行性和无法治愈的神经退行性疾病，具有较长的临床前期，因而对AD病人的早期诊断显得尤为重要。AD发生不是单个基因异常的结果，而是大脑内多个区域的细胞间或细胞内相互影响的结果。因而利用基于网络的分析方法研究AD病人的转录组数据，可以系统地探究AD在分子水平上的复杂性。本研究搜集了3组AD相关的转录组数据并进行预处理，通过WGCNA（Weighted gene co-expression network analysis）方法分别在AD病人和健康对照组中构建了16个和29个共表达基因网络。为了比较健康对照组和AD病人的共表达网络之间的差异，将健康对照组的所有共表达基因网络作为参考网络，AD病人的所有共表达基因网络视为测试网络，查看在测试网络中有多少参考网络被保留了下来，有多少参考网络不在测试网络中。我们发现健康对照组中的橙色共表达网络模块在AD病人的共表达网络中是一种很少被保留的状态，即健康对照组中橙色模块内基因所对应的共表达模式在AD病人中发生了很大的改变。因而对橙色共表达网络模块内的33个基因进行了细胞组分、通路富集和拓扑关系分析。发现这33个基因大部分都与细胞内的线粒体组分有关；被富集到的KEGG通路有帕金森氏病、氧化磷酸化、阿尔茨海默症和亨廷顿舞蹈病等，富集结果也说明健康对照组的橙色共表达模块确实与AD的发生有着密切的关系。通过网络节点的中心性（即通过给定节点的最短路径的数量和给定节点是否处于网络的中心位置），对橙色共表达网络的拓扑分析发现有10个基因（AMZ2、TSPAN3、TMEM150C、FIBP、PHB2、MRPL16、CAPZA2、CETN2、COX6B1和LDHB）在两个网络内的中心性显著不同，在各自网络内拓扑位置截然不同，因而在各自的网络中发挥不同的生物学作用。3. 基于化学基因组学－随机森林模型的中草药抗中风活性化合物的互作靶点发掘中风是高发病率和死亡率的一种突发性脑血管疾病，若中风病人没有得到及时的救治，发生损伤的脑区将会产生永久的功能障碍。在我国使用中草药治疗中风有着悠久的历史，但是中草药植物所含的化合物成分比较复杂，在人体中与这些中草药抗中风的活性化合物有相互作用的蛋白质分子究竟有多少及其作用机制都不是很清楚。本研究基于化学基因组学－随机森林方法，以ChEMBL数据库中已有的化合物－蛋白质相互作用的数据作为训练集，构建稳定的预测模型。将治疗中风的中草药相关的活性化合物作为测试集进行筛选，预测获得了人体内可能的与这些中草药化合物有作用的蛋白质靶点1320个，其中有79个蛋白靶点是已知的治疗中风的药物靶点。通过模型预测获得的1320个蛋白靶点中，与100种以上的治疗中风中草药化合物有相互作用的蛋白靶点为588个，联合分析中风病人的差异表达数据，发现其中有17个基因差异表达。这17个基因中有12个基因的表达产物是酶蛋白；参与细胞内多种生物过程；有2个基因（KLF5、JUN）的表达产物是转录因子；另外的3个基因分别与炎症反应、G蛋白偶联受体和微管组成相关；同时有1个基因（PTGS2）是已知的治疗中风药物的靶点，对应的药物是乙酰水杨酸。因此，除了基因PTGS2以外，预测得到的16个蛋白靶点与上百种治疗中风中草药相关的化合物有相互作用，且在中风病人中显著的差异表达，它们有可能成为治疗中风新的蛋白靶点。

Transcriptome is the sum of all transcript RNAs in a tissue or cell at a specific time and space. Transcription is the first step and play a key regulation role in gene expression. Genes expression can transform genetic information to identifiable phenotypic characteristics. Transcriptomics is the study of gene expression from the RNA level in cells and is an effective means to explore the phenotype of cells and the function of biomolecules. With the development of second-generation sequencing technology, transcriptome data of various human diseases or physiological conditions are accumulating. Using different data-mining methods to explore normal biological processes and pathological mechanisms of diseases from gene expression levels has become a hot and difficult point of research. In order to better resolve complex biological problems, it is necessary to integrated analyze the massive transcriptome data. Reasonable and effective methods or algorithms are used to combine multiple independent research datasets of a certain disease or physiological state. It can effectively expand the sample size and improve the credibility and consistency of the analysis results. Therefore, this study around the transcriptome data by integrating other different types of data and different data-mining methods to explore two complex diseases (Alzheimer's disease and stroke) common in the elderly population. The following is the three relatively independent studies and have made some progress.1. Low expression of the aging-related gene NRXN3 increases the risk of Alzheimer's diseaseAlzheimer's disease (AD) is a common neurodegenerative disease, and aging is considered to be the greatest risk factor for AD. The older the population, the higher the incidence of AD. The purpose of this study was to investigate how aging leads to the increased risk of AD, to find common and concurrent changes in normal aging and AD patients, to find genes that are aging-dependent and promote AD development and progression. We collected 3 groups of AD and 1 group of aging-related expression datasets. And we divided the aging dataset into youth, middle-aged and elderly groups. The two types of expression datasets were analyzed by different algorithms. The MetaMA method was used to calculate the differential expression genes in AD patients and healthy controls. The SAM method was used to identify genes closely related to age (senescence) in the aging expression dataset, including genes positively and negatively correlated with age. This study found that among the differentially expressed genes in AD patients, 1291 genes are closely related to aging. Basically, genes that are up-regulated in AD patients are positively correlated with aging, and genes that are down-regulated in AD patients are negatively correlated with the aging. That is the expression tendency of these genes is basically consistent in AD patients and healthy elderly populations. Enrichment analysis of the 1291 genes in the KEGG pathway, most of which are enriched in KEGG pathways such as neurodegenerative diseases, apoptosis, Alzheimer's disease, and ubiquitin-mediated proteolysis, indicating these genes increased risk of AD in normal aging individuals. By integrating single nucleotide polymorphisms of genes associated with the pathogenesis of AD, it was found that 32 of the 1291 dysregulated genes carrying a single nucleotide polymorphism associated with the pathogenesis of AD. Among them, the gene NRXN3 is the second differentially expressed (down-regulated) gene in AD patients, and NRXN3 is negatively correlated with aging. We focused on the functional network of NRXN3 and its related genes. We found that NRXN3 plays an important role in synaptic function and neural activity. The down-regulation of NRXN3 in healthy aging population is an important genetic risk factor for AD. 2. Exploring potential pathogenic genes in Alzheimer's disease through a co-expression network analysisAlzheimer's disease (AD) is a progressive and incurable neurodegenerative disease with a long preclinical stage. It is particularly important to early diagnosis of the AD patients. The occurrence of AD is not the result of a single genetic abnormality, but the result of intercellular or intracellular interactions in multiple regions of the brain. Therefore, using network-based analysis methods to study the transcriptome data of the AD patients, we can systematically explore the complexity of AD at the biological molecular level. In this study, three groups of the AD-related transcriptome data were collected and preprocessed. In AD patients and healthy controls, 16 and 29 co-expressed genes networks were respectively constructed by WGCNA (Weighted gene co-expression network analysis). In order to compare the differences between the co-expression networks of healthy controls and AD patients, all co-expressed genes networks of healthy controls were used as reference networks, and all co-expressed genes networks of AD patients were considered as test networks. To see how many reference networks were retained in the test network and how many reference networks were not in the test networks. We found that the orange co-expression network module in the healthy control group was rarely retained in the co-expression networks of AD patients. That is, the co-expression pattern corresponding to the genes in the orange module of the healthy control group has undergone a great change in AD patients. Therefore, the cellular components and pathway enrichment analysis were performed on 33 genes in the orange co-expression network module of the healthy controls. It was found that most of these 33 genes are related to mitochondrial components of the cells. The enriched KEGG pathway includes Parkinson's disease, oxidative phosphorylation, Alzheimer's disease and Huntington's disease. The enrichment results also indicated that the orange co-expression module of the healthy control group was indeed closely related to the occurrence of AD. Through the betweenness centrality of the network nodes (i.e., the number of shortest paths for a given node and whether a given node is the center role of the network), the topological analysis of the orange co-expression network found that 10 genes (AMZ2, TSPAN3, TMEM150C, FIBP, PHB2, MRPL16, CAPZA2, CETN2, COX6B1 and LDHB) are significantly different in the topology structure of the betweenness centrality in the two networks. The topological positions of these ten genes are distinct different and thus they play different biological roles in their respective networks.3. Based on chemical genomics-random forest model to identify proteins which can interacted to the Chinese herbal medicine anti-stroke active compoundsStroke is a sudden cerebrovascular disease with high morbidity and mortality. If there is no immediate treatment to the stroke patients, the damaged brain area will have permanent dysfunction. In China, the usage of the Chinese herbal medicine for stroke therapy has a long history, but the composition of the compounds contained in the Chinese herbal medicine plants is relatively complicated. The protein molecules and mechanism of interaction in the human body interacting with the active compounds of these Chinese herbal medicines against stroke are not very clear. The number of protein molecules that interact with these herbal anti-stroke active compounds in humans and their mechanisms of interactions are not well understood. Based on the chemical genomics-random forest method, this study used the existing compound-protein interaction data in the ChEMBL database as a training dataset to construct a stable prediction model. The Chinese herbal medicine-related active compounds for stroke treatment were screened as test datasets. And 1320 protein targets interacting with these herbal compounds were obtained by model prediction, of which 79 proteins were the known drugs' targets for the stroke treatment. There were 588 protein targets interacting with more than 100 therapeutic stroke herbal compounds. Moreover, genes differential expression of the stroke patients was analyzed. It was found that 17 genes of the 1320 predicted targets were differentially expressed. Among the 17 genes, the expression products of 12 genes are enzyme proteins involved in various biological processes in the cell; two genes (KLF5, JUN) are transcription factors, and the other three genes are related to inflammatory, G protein coupled receptor and microtubule composition. At the same time, one gene (PTGS2) is a known drug target of stroke treatment and the corresponding drug is acetylsalicylic acid. Therefore, in addition to the gene PTGS2, the 16 predicted proteins targets can interact with hundreds of the stroke-related Chinese herbal medicine compounds and they are significantly differentially expressed in stroke patients. So, they may become new protein targets for the treatment of stroke.