| 引用本文: | 魏丕静,刘晶晶,赵永敏,苏延森,郑春厚.基于多层基因网络的关键基因识别算法[J].生物信息学,2023,21(4):277-285. |
| WEI Pijing,LIU Jingjing,ZHAO Yongmin,SU Yansen,ZHENG Chunhou.Key gene identification algorithm based on multi-layer network[J].Chinese Journal of Bioinformatics,2023,21(4):277-285. |
|
| 摘要: |
| 疾病关键基因可用于疾病诊断、预测和新药或新疗法有效性的评价,故识别与疾病紧密相关的关键基因十分重要。然而现在有些疾病样本数据较少,传统基于大样本的关键基因挖掘方法不适用于该类数据。本文针对含少量样本数据的疾病,首先利用单样本网络构建方法构建每个疾病样本的个体化基因网络,并通过建立基因间的层间联系构建多层基因网络。然后利用基于张量的多层网络中心性方法评估每层网络中基因间的相互作用以及层间影响,对基因进行重要性打分,识别疾病关键基因。最后将该方法应用到哮喘数据集上,并与经典算法进行比较,结果表明,利用该方法所识别的已获批准的药物靶标基因的排名较优；对所得到的新的潜在关键基因TP53、PUS10、MAP3K1等进行功能和通路富集分析,结果表明其与哮喘有紧密关联。 |
| 关键词: 多层基因网络 随机游走 节点中心性 关键基因 |
| DOI：10.12113/202204003 |
| 分类号:Q343.1 |
| 文献标识码:A |
| 基金项目: |
|
| Key gene identification algorithm based on multi-layer network |
|
WEI Pijing1, LIU Jingjing2, ZHAO Yongmin2, SU Yansen3,ZHENG Chunhou3
|
|
(1. Institutes of Physical Science and Information Technology, Anhui University, Hefei 230601, China; 2. School of Computer Science and Technology, Anhui University, Hefei 230601, China;3. School of Artificial Intelligence, Anhui University, Hefei 230601, China)
|
| Abstract: |
| Critical genes of diseases can be used to diagnose diseases, predict and evaluate the effectiveness of new drugs or new therapies, so it is very important to identify critical genes closely related to diseases. However, the samples of some diseases are limited. It is difficult to apply the traditional methods based on large sample data to mine critical genes for these diseases. In this paper, for diseases with small amount of samples, we first construct sample-specific network for each sample with the single-sample network constructing methods,and construct a multi-layer gene network by estabishing inter-layer connections between genes. A tensor-based multi-layer network centrality approach is then used to assess the interactions between genes in each layer of the network and the inter-layer effects to score the genes for importance and identify disease key genes. Finally, the method is used to two asthma datasets and compared with the classical algorithm. The results show that compared with other methods, the approved drug target genes rank higher in the gene rankings obtained by this method. And function and pathway enrichment analysis of the new potential critical genes TP53, PUS10, MAP3K1,etc. indicat that they were closely related to asthma. |
| Key words: Multilayer gene interaction network Random walk Node centrality Critical genes |
