A Method of Motif Mining Based on Backtracking and Dynamic Programming

Because of the complexity of biological networks, motif mining is a key problem in data analysis for such networks. Researchers have investigated many algorithms aimed at improving the efficiency of motif mining. Here we propose a new algorithm for motif mining that is based on dynamic programming and backtracking. In our method, firstly, we enumerate all of the 3-vertex sub graphs by the method ESU, and then we enumerate sub graphs of other sizes using dynamic programming for reducing the search time. In addition, we have also improved the backtracking application in searching sub graphs, and the improved backtracking can help us search sub graphs more roundly. Comparisons with other algorithms demonstrate that our algorithm yields faster and more accurate detection of motifs.