Ranking Beta Sheet Topologies of Proteins

One of the challenges of protein structure prediction is to identify long-range interactions between amino acids. To reliably predict such interactions, we enumerate, score and rank all βtopologies (partitions of β-strands into sheets, orderings of strands within sheets and orientations of paired strands) of a given protein. We show that the β-topology corresponding to the native structure is, with high probability, among the top-ranked. Since full enumeration is very time-consuming, we also suggest a method to deal with proteins with many βstrands. The results reported in this paper are highly relevant for ab initio protein structure prediction methods based on decoy generation. The top-ranked βtopologies can be used to find initial conformations from which conformational searches can be started. They can also be used to filter decoys by removing those with poorly assembled β-sheets, and finally they can be relevant in contact prediction methods.