Distributed Parallel Needleman-Wunsch Algorithm on Heterogeneous Cluster System

At first, the needleman-wunsch algorithm that relies the optimal matching result is regarded as the basic algorithm to solve the problem of sequence alignment. Nevertheless, as the result of sequence database capacity index increased year by year and the algorithm of time and space complexity are higher, O (m*n). Facing these problems, researchers need to reduce the sequence alignment algorithm becomes more important for time and space complexity basic algorithm to solve the problem of sequence alignment. Therefore, this paperuses the characteristics of heterogeneous cluster , basing on the theory of divisible load, using the communication strategy of LIFO, simplex communication model, the design bases on dual sequence global alignment parallel algorithm of Needleman-Wunsch, the algorithm puts forward the strategy of partitioning score data matrix , to determine the optimal number of iterations and assigns to the child nodes of sub-sequence length ability to making full use of each node and performs tasks in a row, at the same time puts forward the strategy of backtracking, reduces the algorithm operation time, makes integral algorithm optimal.