Sequence Order Independent Comparison of Protein Global Backbone Structures and Local Binding Surfaces for Evolutionary and Functional Inference

Alignment of protein structures can help to infer protein functions and AQ1 can reveal ancient evolutionary relationship. We discuss computational methods we developed for structural alignment of both global backbones and local surfaces of proteins that do not depend on the ordering of residues in the primary sequences. The algorithm for global structural alignment is based on fragment assembly, and takes advantage of an approximation algorithm for solving the maximum weight independent set problem. We show how this algorithm can be applied to discover proteins related by complex topological rearrangement, including circularly permuted proteins as well as proteins related by complex higher order permutations. The algorithm for local surface alignment is based on solving the bi-partite graph matching problem through comparison of surface pockets and voids, such as those computed from the underlying alpha complex of the protein structure. We also describe how multiple matched surfaces can be used to automatically generate signature pockets and basis set that represents the ensemble of conformations of protein binding surfaces with a specific biological function of binding activity. This is followed by illustrative examples of signature pockets and basis set computed for NAD binding proteins, along with a discussion on how they can be used for discriminating NAD-binding enzymes from other enzymes. Introduction To understand the molecular basis of cellular processes, it is important to gain a comprehensive understanding of the biological functions of protein molecules. Although an increasing number of sequences and structures of proteins become available, there are many proteins whose biological functions are not known, or knowledge of their biological roles is incomplete. This is evidenced by the existence J. Dundas (B) Bioinformatics Program, Department of Bioengineering, University of Illinois Chicago, Wolcott, Chicago, IL 60612-7340, USA e-mail: [email protected] D. Kihara (ed.), Protein Function Prediction for Omics Era, DOI 10.1007/978-94-007-0881-5_7, C © Springer Science+Business Media B.V. 2011 U N C O R R E C TE D P R O O F SPB-183939 Chapter ID 7 February 21, 2011 Time: 08:05pm Proof 1