A hierarchical link clustering based approach for identifying protein complexes by incorporating core-attachment structure

摘要

Determining how to identify protein complexes automatically and effectively from experimental datasets is a challenging and meaningful work in proteomics and bioinfor-matics. In this paper, we propose a simple but effective method, called HLC-CA, to predict protein complexes by making full use of the inherent core-attachment structure of complexes. First, HLC-CA obtains candidate clusters by employing an appropriate hierarchical link clustering algorithm. Secondly, it filters the clusters to identify complex cores by adopting simple criteria. Thirdly, it recruits attachments for each core by using a topological feature. Finally, it composes the cores and attachments to form protein complexes. Extensive experimental results show that HLC-CA significantly outperforms the state-of-the-art methods.