Solving the protein threading problem by Lagrangian relaxation

This paper presents an efficient algorithm for aligning a query amino-acid sequence to a protein 3D structure template. Solving this problem is one of the main steps of the methods of protein structure prediction by threading. We propose an integer programming model and solve it by branch-and-bound algorithm. The bounds are computed using a Lagrangian dual of the model which turns out to be much easier to solve than its linear programming relaxation. The Lagrangian relaxations are computed using a dynamic programming algorithm. The experimental results show that our algorithm outperforms the commonly used methods. The proposed algorithm is general enough and can be easily plugged in most of the threading tools in order to increase their performance.