Abstract: The selection of the optimal seed(that is， the seed with the lowest frequency)is a complex calculation problem， which often takes a long time. A frequency-based merge seed selection(FMSS) algorithm is proposed， which can efficiently select the suboptimal set of seeds and improve the performance of existing mapping tools. In the experiment， FMSS was compared with the average seed selection method and the optimal seed solver(OSS). Experimental results show that FMSS can select the optimal set of seeds close to OSS， and the time cost of FMSS is far lower than that of the OSS algorithm. The FMSS algorithm is more suitable for seed selection in terms of time cost and seed selection quality.

Key words: seed frequency, read-mapping, average seed;frequency merge, optimal seed