Abstract: The optimal sequence distribution models for solving Motif finding with length L镇16 and L } 16 were cons tructed respectively and a parallel algorithm to find Motif combining voting algorithm with uniform projection and neighbourhood thresholding algorithm was implemented on the heterogeneous cluster that the processor nodes have dif- ferent computing speed and distinct communication capability. Experimental results show that the parallel Motif finding algorithm using optimal sequence distribution strategy has good speedup and scalability, and it is superior to the parallel algorithm using even distribution strategy.

Key words: DNA sequence, Motif finding, Parallel algorithm, Heterogeneous clusters, Data distribution model