Mining Weighted Protein Complexes Based on Fuzzy Ant Colony Clustering Algorithm

• Published in: Smart City and Informatization pp. 557 - 569
• Main Authors: Mao, Yimin, Deng, Qianhu, Liu, Yinping
• Format: Book Chapter
• Language: English
• Published: Singapore Springer Singapore
• Series: Communications in Computer and Information Science
• Subjects: Ant colony clustering algorithm; FCM; Fitness; Protein complex; Protein-protein interaction (PPI) network
• ISBN: 9789811513008; 9811513007
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-981-15-1301-5_44

Aiming at the defect that the accuracy of the protein complexes based on fuzzy ant colony clustering is not high, the time performance and recall are low, a novel algorithm named FAC-PC (mining weighted protein complexes based on fuzzy ant colony clustering algorithm) is proposed. The weighted protein network is established by the integration of both edge aggregation coefficient and gene expression data to eliminate the effect of false positives, and the selection of essential protein by using a new function EPS (essential protein selection). Then, this paper proposes that PFC (protein fitness calculation) and SI (similarity improvement) overcome the problems of massive merger, repeated picking and dropping operations in ant colony clustering algorithm. Furthermore, a new FCM (fuzzy C-means) objective function which takes a balance between inter-clustering and intra-clustering variation is proposed for protein complexes. The experimental results show that the superiority of the FAC-PC algorithm in terms of accuracy and computational time.