Peng Jiaxin, Kuang Linai, Zhang Zhen, Tan Yihong, Chen Zhiping, Wang Lei
College of Computer, Xiangtan University, Xiangtan, China.
College of Computer Engineering and Applied Mathematics, Changsha University, Changsha, China.
Front Genet. 2021 Jul 29;12:721486. doi: 10.3389/fgene.2021.721486. eCollection 2021.
In recent years, many computational models have been designed to detect essential proteins based on protein-protein interaction (PPI) networks. However, due to the incompleteness of PPI networks, the prediction accuracy of these models is still not satisfactory. In this manuscript, a novel key target convergence sets based prediction model (KTCSPM) is proposed to identify essential proteins. In KTCSPM, a weighted PPI network and a weighted (Domain-Domain Interaction) network are constructed first based on known PPIs and PDIs downloaded from benchmark databases. And then, by integrating these two kinds of networks, a novel weighted PDI network is built. Next, through assigning a unique key target convergence set (KTCS) for each node in the weighted PDI network, an improved method based on the random walk with restart is designed to identify essential proteins. Finally, in order to evaluate the predictive effects of KTCSPM, it is compared with 12 competitive state-of-the-art models, and experimental results show that KTCSPM can achieve better prediction accuracy. Considering the satisfactory predictive performance achieved by KTCSPM, it indicates that KTCSPM might be a good supplement to the future research on prediction of essential proteins.
近年来，许多计算模型被设计用于基于蛋白质 - 蛋白质相互作用（PPI）网络来检测必需蛋白质。然而，由于PPI网络的不完整性，这些模型的预测准确性仍然不尽人意。在本论文中，提出了一种基于关键靶标收敛集的新型预测模型（KTCSPM）来识别必需蛋白质。在KTCSPM中，首先基于从基准数据库下载的已知PPI和PDI构建一个加权PPI网络和一个加权（结构域 - 结构域相互作用）网络。然后，通过整合这两种网络，构建一个新型加权PDI网络。接下来，通过为加权PDI网络中的每个节点分配一个唯一的关键靶标收敛集（KTCS），设计一种基于带重启的随机游走的改进方法来识别必需蛋白质。最后，为了评估KTCSPM的预测效果，将其与12种具有竞争力 的最新模型进行比较，实验结果表明KTCSPM能够实现更好的预测准确性。考虑到KTCSPM所取得的令人满意的预测性能，这表明KTCSPM可能是未来必需蛋白质预测研究的一个很好的补充。
