关键词: DNA计算, 层次聚类, 最小生成树, 粘贴模型, 2-臂DNA分子

Abstract: In order to take full advantage of the high parallelism and huge storage capacity of DNA molecules in biological computing, this paper introduced DNA computing into hierarchical clustering to do global research on data set. For realizing the nearest neighbor hierarchical clustering, an algorithm combining sticker model with 2-armed DNA molecules was put forward. Based on the idea of MST (Minimum Spanning Tree), the first thing to do was generating complex DNA strands of all combinations of edges and then screening those containing n-1 edges. Based on the edges, it is needed to set the corresponding vertex stickers and keep those strands covering all the vertices. After that, weight strands constructed by 2-armed molecules would be appended at the end of the complex strands and the shortest ones could be detected by gel electrophoresis. Finally, by fluorescence analysis the clustering result can be got. In computer simulation, this algorithm may take different lengths of edges into account instead of varying the polynomial time complexity and the number of steps to read final results is set as a constant.

Key words: DNA computing, hierarchical clustering, Minimum Spanning Tree (MST), sticker model, 2-armed DNA molecule