Title :
Solving the contact map overlap problem via tree decomposition and a DEE-like pruning strategy
Author_Institution :
Toyota Technol. Inst., Chicago
Abstract :
This paper proposes a new algorithm for protein structure alignment in the case where a protein structure is represented by a contact map graph and the sequential order of residues is strictly enforced in the alignment. The time complexity of the algorithm, O(n(2n+w)!/((2n)!w!)) where n is the maximum number of residues in two proteins and w = O((Du/Dl) n2/3 log n) is the treewidth of a contact graph. In particular, Du is the contact distance cutoff and Dc is the minimum inter-residue distance in a typical protein. If the distance between two matched residues after two proteins are superimposed is bounded by a parameter Dc, then the time complexity is O((n/epsivDc)6((n+w)!/(n!w!)Deltaw)) where e is a small constant and Delta = (1 + (2Dc/Dl))3. This algorithm consists of two major components: tree-decomposition and a DEE-like pruning strategy. The tree-decomposition method exploits the geometric feature of a protein structure and guarantees to terminate within subexponential time. The DEE-like pruning strategy exploits the sequential constraint and can quickly eliminate non-optimal alignments from a large search space. Based on these two techniques, we have developed a structure alignment program TreeAlign to find the globally optimal solution of the contact map overlap problem. The preliminary experimental results indicate that on a Linux PC with 2.5 GHz CPU, TreeAlign can align two similar proteins within seconds and two dissimilar proteins within minutes.
Keywords :
biology computing; graph theory; macromolecules; molecular configurations; molecular orientation; proteins; CPU; DEE-like Pruning Strategy; Linux PC; TreeAlign; contact map graph; contact map overlap problem; geometric feature; protein structure alignment; sequential constraint; tree decomposition; Computational complexity; Computational efficiency; Instruments; Lagrangian functions; Linux; Polynomials; Proteins; Tree graphs; USA Councils; DEE (Dead-End Elimination); branch-and-bound; contact map overlap; tree-decomposition;
Conference_Titel :
Decision and Control, 2007 46th IEEE Conference on
Conference_Location :
New Orleans, LA
Print_ISBN :
978-1-4244-1497-0
Electronic_ISBN :
0191-2216
DOI :
10.1109/CDC.2007.4434352