Institute of Electrical & Electronic Engineers
Understanding which characteristics of proteins have most impact on their functional role is one of the main challenges of the post-genomic era. Surface-based techniques for protein comparison and classification typically require a compact surface representation, capable of effectively condensing its description. In this paper, the authors propose an original template-matching algorithm for multi-feature surface clustering in the biochemical context. The effectiveness of their clustering algorithm in capturing surface similarities is then discussed within a larger framework for protein classification based on surface comparison, with the support of tests performed on a dataset including 25 proteins.