Discovering important proteins in Protein-Protein Interaction (PPI) networks has attracted a lot of attention in recent years. Most of the previous work applies different network centrality measures such as Closeness, Betweenness, PageRank and many others to discover the most influential proteins in PPI networks. Although entropy is a well-known graph-based method in computer science, according to our knowledge, it is not used in the biology domain for this purpose. In this paper, first, we annotate the human PPI network with available annotation data. Second, we introduce a new concept called annotation-context that describes each protein according to annotation data of its neighbors. Third, we apply an entropy measure to discover proteins with varied annotation-context. Empirical results indicate that our proposed method succeeded in (1) differentiating essential and non-essential proteins in PPI networks with annotation data; (2) outperforming centrality measures in the task of discovering essential nodes; (3) predicting new annotated proteins based on existing annotation data.

中文翻译：

---

EMDIP：一种熵测度，用于发现PPI网络中的重要蛋白质。

近年来，在蛋白质-蛋白质相互作用（PPI）网络中发现重要的蛋白质引起了很多关注。以前的大多数工作都采用了不同的网络中心性度量，例如“紧密度”，“中间性”，“ PageRank”和许多其他方法，以发现PPI网络中最具影响力的蛋白质。尽管熵是计算机科学中众所周知的基于图的方法，但据我们所知，它并未在生物学领域中用于此目的。在本文中，首先，我们使用可用的注释数据对人类PPI网络进行注释。其次，我们引入了一个称为注释上下文的新概念，该概念根据其邻居的注释数据来描述每种蛋白质。第三，我们应用熵测度来发现具有不同注释环境的蛋白质。实验结果表明，我们的方法成功地实现了（1）利用注释数据区分PPI网络中的必需蛋白和非必需蛋白。（2）在发现基本节点的任务中胜过中心化措施；（3）根据现有注释数据预测新的注释蛋白质。