Protein structure abstraction and automatic clustering using secondary structure element sequences

To study protein clustering is very important in diverse fields such as drug design and environmental industry. For a meaningful clustering, protein structure must be considered. But, protein structures are very complicated and have so much information such as angles, 3-dimensional coordinates. Thus, it is not easy to efficiently compute their relations. In this paper, we present a method to efficiently abstract and cluster protein structures using secondary structure element sequences. Since a secondary structure element sequence is an abstract representation of protein structure, it can be regarded as a useful descriptor to cluster a set of proteins at the abstraction level. Using secondary structure element sequences and their distances, we implemented an automatic protein clustering system and verify their efficiency by experimental results.