A New Feature in Protein-Protein Interaction Network to Identify Essential Genes
Essential genes are responsible for an organisms viability, and its an important issue to identify essential genes in systems biology. Many studies have found that there is a close relationship between gene essentiality and network topology in protein-protein interaction network (PPIN). It is indicated that nodes with high centralities in PPIN tend to be essential. Well-known network centrality measures have always been used by investigators to predict essential genes with machine learning methods. However, the prediction performance of essential genes by these centrality measures is not satisfactory. Importing new features in PPIN which correlate with gene essentiality is a valuable work to further improve the predictability of essential genes. In this work, we defined a new feature called AT-neighbors essentiality (KNE) which reflects the clustering ability of essential nodes in PPIN. This measure was proved to be closely related with gene essentiality in PPIN that genes with high KNE value tend to be essential. Besides, KNE shows better correlation with gene essentiality than common centrality measures. Moreover, by using a random forest classifier, we were able to demonstrate that KNE can act as a better predictor than common centrality measures in the identification of essential genes (AUC value increased by about 3％).
essential genes protein-protein interaction network network centrality K-neighbors essentiality
Hanbin Wang Lipeng Wang
College of Computer Science, Sichuan University Chengdu, Sichuan, China
国际会议
2011 International Conference on Database and Data Mining(ICDDM 2011)(2011年数据库和数据挖掘国际会议)
三亚
英文
50-54
2011-03-25（万方平台首次上网日期，不代表论文的发表时间）