Speedup LP Approach to Protein Threading via Graph Reduction
In our protein structure prediction computer program RAPTOR, we have implemented a linear programming (LP) approach to protein threading based on the template contact map graph. Our protein threading model considers pairwise contact potential rigorously. In order to further improve the computational efficiency of our LP approach, this paper proposes a graph reduction technique to reduce a template contact graph to a new one with fewer vertices and edges. Our graph reduction operation is formalized to minimize the number of variables, constraints and non-zero elements in the constraint matrix of our linear programs. These three factors are key to the computational time of solving linear programs by both the Simplex method and the Interior-Point method. This graph reduction technique does not impact the quality of protein threading (i.e., the energy function is still globally optimized). Experiments show that the more a template contact graph can be reduced, the more computational efficiency improvement can be attained and that in average, the computational efficiency of threading any long sequence to the whole template database can be improved by 30%.
Unable to display preview. Download preview PDF.
- 1.Xu, J., Li, M., Lin, G., Kim, D., Xu, Y.: Protein threading by linear programming. In: Biocomputing: Proceedings of the 2003 Pacific Symposium, Hawaii, USA, pp. 264–275 (2003)Google Scholar
- 3.Fischer, D.: (December 2002), http://www.cs.bgu.ac.il/~dfischer/CAFASP3/
- 9.Yanev, N., Andonov, R.: Protein threading is in P? Technical Report INRIA RR-4577, French (October 2002)Google Scholar
- 10.Lathrop, R.H., Smith, T.F.: A branch-and-bound algorithm for optimal protein threading with pairwise (contact potential) amino acid interactions. IEEE Computer Society Press, Los Alamitos (1994)Google Scholar
- 11.CASP5 (December 2002), http://predictioncenter.llnl.gov/casp5/Casp5.html
- 12.Rychlewski, L. (2002), http://www.bioinfo.pl/livebench/6