衰老基因和癌基因的特征分析与预测

发布时间：2018-03-21 21:01

  本文选题：衰老基因　切入点：癌基因　出处：《电子科技大学》2011年博士论文　论文类型：学位论文

【摘要】：衰老和癌是密切相关的生命现象:癌的发生率随衰老而上升。寻找衰老基因和癌基因是阐述两者关系的基础,也是当前该领域的重要工作。因为湿实验筛选衰老基因和癌基因非常费时费力,如何分析高通量生物技术产生的数据并发展算法来帮助生物学家减少实验工作量就显得十分重要。在本工作中,我们首次系统地分析了衰老基因的特征,在此基础上,提出了一个预测衰老基因的算法;进一步,我们通过生物实验证实了一些预测的结果;针对最新的癌样本突变组数据,我们发展了一个简单高效的癌基因预测算法;对于当前很多癌基因功能注释比较粗糙的问题,我们发展了一个预测癌基因精细功能的算法。最后,基于蛋白质相互作用网络,我们分析并发现了衰老基因和癌基因的一些共同特征,这些特征较好地解释了衰老和癌的密切关系。 详细地,本文主要包括以下五部分内容: 1.衰老基因特征分析与预测。与非衰老基因相比较,我们系统地分析了衰老基因的特征。发现衰老基因倾向于(1)有更长的基因和蛋白序列,(2)和其它基因有更高的表达相关性,(3)在某些功能和表形上明显地聚集,(4)更高的序列保守性,(5)位于蛋白质相互作用网络的中心等网络拓扑特征。基于这些特征,我们发展了一个基于支持向量机的衰老基因预测算法。利用该算法,以高于0.85的精确率预测了243个新的衰老基因。进一步,我们评估了各个特征对预测结果的贡献,衰老基因所富集的功能贡献最大而其所富集的表型次之。 2.实验验证预测的衰老基因。对于预测的衰老基因,从文献和实验两个角度进行了证实。文献角度,一些预测的基因已被其它实验室证明参与寿命调节,例如,vps-34被算法预测为衰老基因,有研究组报道vps-34对于饮食限制引起的寿命延长是必须的。实验角度,从预测的衰老基因列表中,我们选择并沉默了7个基因观察其对寿命的影响,发现沉默基因B0025.1或者F58F12.1的表达可以延长daf-2突变体的寿命而沉默F54C9.1则缩短了daf-2突变体的寿命。 3.癌基因的特征分析与预测。针对高通量的癌基因组突变谱数据,我们提出了一个基于癌功能类预测癌基因的算法。结合癌功能类与基因的非同义突变个数,我们的算法在预测的准确性上要明显优于前人基于选择压力和非同义突变个数的算法。最后,应用算法,我们将46个注释到癌功能类并且非同义突变个数至少为3的激酶基因预测为癌基因,这对进一步的生物实验可能有帮助。 4.癌基因精细功能预测。目前,癌基因功能和癌信号通路知识是有限的且粗糙的,成为癌机制研究的一个瓶颈。具体地,很多癌基因只是注释到GO数据库的高层功能类。在这里,我们开发了一个基于功能特异的蛋白质相互作用网络的高效算法来寻找癌基因的细致功能。通过利用蛋白质已知的功能知识,193个癌基因被预测到了更细致的功能类。进一步,我们选择并定义了一组癌功能类,应用算法,221个基因被预测到了这组功能类,提高了功能特异的相互作用子网的连通性,使得对癌功能类的认识更为清晰和完整。 5.衰老基因和癌基因的关系。针对衰老和癌表型上的密切关系,利用已有的衰老基因、癌基因和蛋白质相互作用数据,从三方面进行了分析,我们发现(1)衰老基因和癌基因集合高度重合,很多衰老基因本身就是癌基因;(2)在蛋白质相互作用网络上,衰老基因和癌基因有着相似的网络拓扑特征,都倾向位于网络的中心,对网络全局起着调控作用。(3)衰老基因和癌基因它们自己以及之间倾向直接的相互作用。这些结果都较好的解释了衰老和癌的密切关系。 综上,在分析总结衰老基因和癌基因特征的基础上,我们分别提出了预测衰老基因、癌基因和癌基因精细功能的算法,并对衰老基因和癌基因的共同特征进行了分析。一些预测的结果已经被实验证实。这将推动衰老和癌的研究。
[Abstract]:Aging and cancer is closely related to the phenomenon of life: cancer incidence increased with aging. In search of aging genes and oncogenes is discussed the relationship between the two, is an important work in the field at present. Because the wet screening experiment of aging genes and oncogenes is very time-consuming, how to analyze the high flux generated data and biological technology the development of algorithms to help biologists seem to decrease the workload is very important. In this work, we systematically analyzed the characteristics of aging gene, on this basis, this paper proposes a prediction algorithm of aging gene; further, we confirmed some prediction results through biological experiments; the latest cancer samples mutation group the data, we develop a simple and efficient algorithm for the prediction of cancer gene; gene function of many cancer notes is relatively rough, we develop a prediction of cancer Finally, based on the protein interaction network, we analyzed and discovered some common characteristics of aging genes and oncogenes, which better explained the close relationship between aging and cancer.
In detail, this article mainly includes the following five parts:
Analysis and prediction of genetic characteristics of 1. aging. Compared with the non aging gene, we systematically analyzed the characteristics of aging genes. Aging genes tend to (1) gene and protein sequences longer, (2) and other related gene expression is high, (3) in some function and on the table obviously the accumulation of sequence conservation (4) higher, (5) is located in the center of the network topology characteristics of the protein interaction network. Based on these features, we developed an algorithm to predict aging gene based on support vector machine. Using this algorithm, with high accuracy to 0.85 of the predicted 243 new aging gene. Further, we evaluated the contribution of each feature on the prediction results, the aging gene enriched functional contribution and the enrichment of the phenotype.
2. aging gene prediction. Experimental verification for aging gene prediction, from two aspects of literature and experiment were confirmed. The literature point of view, some of the predicted genes have been shown to be involved in regulating other laboratory life, for example, vps-34 algorithm prediction for aging gene, the study group reported that vps-34 is necessary for prolonged dietary restriction caused by life. From the point of view of experimental aging gene prediction list, we selected 7 genes silence and to observe its influence on life, found that silence gene expression of B0025.1 or F58F12.1 can prolong the life of DAF-2 mutant and DAF-2 mutant F54C9.1 silence will shorten life.
Characteristic analysis and prediction of 3. cancer genes. The high-throughput cancer genome mutation spectrum data, we propose an algorithm to predict cancer oncogene function. Based on combination of non synonymous mutations and cancer class number, our algorithm in prediction accuracy is superior to the previous selection based on pressure and the number of non synonymous mutation algorithm. Finally, application of the algorithm, we will have 46 notes to the cancer function class and non synonymous mutations in a number of at least 3 kinase gene prediction for cancer gene, which may be helpful for further biological experiments.
Prediction of 4. cancer gene fine function. At present, gene function and signal pathway of cancer knowledge is limited and rough, become a bottleneck mechanism of cancer research. In particular, many cancer genes are annotated to high-level functional class GO database. Here, we developed an efficient algorithm for functional specificity of protein interactions the role of the network based on the detailed function to find cancer genes. By using knowledge of known protein functions, 193 genes were predicted to cancer features more detailed. Further, we select and define a set of cancer function category, application method, 221 genes were predicted by this set of features, improve the interaction network connectivity function specific, the understanding of cancer function is more clear and complete.
The relationship between the 5. aging genes and oncogenes. According to the close relationship between aging and cancer phenotype, using aging gene the gene and protein interaction data were analyzed from three aspects, we found that (1) aging genes and oncogenes set a high degree of coincidence, many aging gene itself is oncogene; (2) in protein interaction networks, aging genes and oncogenes have similar characteristics of network topology, is located in the center of the network tend, play a role in regulation of the global network. (3) tend to direct interaction of aging genes and oncogenes and their own. These results can explain the close relationship between aging and cancer.
缁间笂,鍦ㄥ垎鏋愭，

本文编号：1645523