Professor Saraswathi Vishveshwara
Molecular Biophysics Unit
Indian Institute of Science
Bangalore, India

Tuesday, March 8, 2005
1:00 pm (CT)
3169 Beckman Institute

Abstract

Protein structures have been viewed as graphs, which have been constructed by considering the non-covalent interactions within the proteins from the available crystal structure data. A distinct advantage of the graph representation is that it takes into account the global topology of the protein structure. The spectral properties of such graphs have been investigated to identify a variety of clusters within proteins, which are important from the structure, stability and folding points of view. Further, sub-clustering of connected graphs has aided in automatic identification of domains and domain-domain interfaces in multi-domain proteins. The network properties such as the hub-propensities, size of the largest cluster have been evaluated as a function of the strength of non-covalent interaction between amino acid residues, from the protein structure graphs. The results are analysed from the perspective of protein structure, function and folding.