Zhou HongFang, Liu Jie, Li JunHuai, Duan WenCong
School of Computer Science and Engineering, Xi'an University of Technology, Xi'an, China.
PLoS One. 2017 Jul 12;12(7):e0180570. doi: 10.1371/journal.pone.0180570. eCollection 2017.
Protein complex detection in PPI networks plays an important role in analyzing biological processes. A new algorithm-DBGPWN-is proposed for predicting complexes in PPI networks. Firstly, a method based on gene ontology is used to measure semantic similarities between interacted proteins, and the similarity values are used as their weights. Then, a density-based graph partitioning algorithm is developed to find clusters in the weighted PPI networks, and the identified ones are considered to be dense and similar. Experimental results demonstrate that our approach achieves good performance as compared with such algorithms as MCL, CMC, MCODE, RNSC, CORE, ClusterOne and FGN.
蛋白质-蛋白质相互作用（PPI）网络中的蛋白质复合物检测在分析生物过程中起着重要作用。提出了一种新算法——DBGPWN，用于预测PPI网络中的复合物。首先，使用基于基因本体的方法来测量相互作用蛋白质之间的语义相似性，并将相似性值用作它们的权重。然后，开发了一种基于密度的图划分算法，以在加权PPI网络中找到簇，并且将识别出的簇视为密集且相似的。实验结果表明，与MCL、CMC、MCODE、RNSC、CORE、ClusterOne和FGN等算法相比，我们的方法具有良好的性能。
