摘要
复杂疾病的发生和发展往往是多种因素共同作用的结果。因此,采用网络模型探索疾病相关生物标志物的研究具有重要意义,但目前还没有一个统一规范的标准去衡量一个网络的构建是否符合真实的生物学特征。基于在较小干扰下的真实生物系统应具有鲁棒性这一生物学假设,论文介绍了一种基于DNA甲基化数据的扰动算法来评估网络稳定性的方法(Evaluation of Network-Construction based on Perturbation,ENCPer),该方法通过对真实数据加入扰动,并比较真实网络和扰动网络的稳定性来评估不同方法构建网络的稳定性。文章从平均值,总体值和差异得分这三种CpG位点融合的角度来构建基因加权网络,使用ENCPer来评估不同方法构建网络的稳定性。通过分析稳定网络的富集结果,发现了潜在的生物学意义。ENCPer还可以帮助研究人员找到更符合真实生物系统的网络,在该网络模型上进行生物分析得出的结果更加真实可信。
The occurrence and development of complex diseases are often the result of the joint action of multiple factors.Therefore,it is of great significance to explore the disease-associated biomarkers based on network model. At present,however,there is no uniform standard for evaluating whether a network is constructed to conform to real biological characteristics. Considering a biological hypothesis that real biological systems for smaller disturbances should be robust,this paper develops an approach,ENCPer,to evaluate the stability of the different network construction methods using perturbation algorithm. This method evaluates the stability of different methods by adding disturbance to real data and comparing the stability of real network and disturbed network. This paper constructs gene weighted network from three perspectives of aggregation of multiple CpG sites:mean,holistic,and differential scores,and evaluates the stability of different networks using ENCPer. Through analyzing the enrichment results of the stable networks,the potential biological significance is found. The ability of ENCPer can help researchers find networks that are more in line with real biological system,and the results of biological analysis based on network model are more reliable.
作者
寇传华
张媛媛
KOU Chuanhua;ZHANG Yuanyuan(School of Information and Control Engineering,Qingdao University of Technology,Qingdao 266520)
出处
《计算机与数字工程》
2022年第12期2746-2751,共6页
Computer & Digital Engineering
