Self-Interacting Proteins (SIPs), whose two or more copies can interact with each other, have significant roles in cellular functions and evolution of Protein Interaction Networks (PINs). Knowing whether a protein can act on itself is important to understand its functions. Previous studies on SIPs have focused on their structures and functions, while their whole properties are less emphasized. Not surprisingly, identifying SIPs is one of the most important works in biomedical research, which will help to understanding the function and mechanism of proteins. It is worth noting that high throughput methods can be used for SIPs prediction, but can be costly, time consuming and challenging. Therefore, it is urgent to design computational models for the identification of SIPs. In this review, the concept and function of SIPs were introduced in detail. We further introduced SIPs data and some excellent computational models that have been designed for SIPs prediction. Specially, the most existing approaches were developed based on machine learning through carrying out different extract feature methods. Finally, we discussed several difficult problems in developing computational models for SIPs prediction.

自相互作用蛋白（SIP），其两个或多个副本可以彼此相互作用，在细胞功能和蛋白相互作用网络（PIN）的进化中具有重要作用。了解蛋白质是否可以对其自身起作用对于了解其功能很重要。先前对SIP的研究主要集中在其结构和功能上，而对它们的整体属性的重视程度则较低。毫不奇怪，鉴定SIP是生物医学研究中最重要的工作之一，这将有助于理解蛋白质的功能和机理。值得注意的是，高吞吐量方法可用于SIP预测，但可能会昂贵，耗时且具有挑战性。因此，迫切需要设计用于识别SIP的计算模型。在这篇综述中，详细介绍了SIP的概念和功能。我们进一步介绍了SIP数据和一些针对SIP预测而设计的出色计算模型。特别是，通过执行不同的提取特征方法，基于机器学习开发了大多数现有方法。最后，我们讨论了在开发用于SIP预测的计算模型时遇到的几个难题。