A local alignment of DNA based on parallelized MUMmer algorithm

Sequence alignment is the foundation and an important part of bioinformatics. In local alignment of DNA, MUMmer Algorithm is a double sequence alignment approximation algorithm based on anchor, which includes the following steps: using the suffix tree to find the set of maximal unique matches(MUM), finding out a group of MUMs with the highest score from the set as the anchors, and finally completing the whole alignment by comparing the gaps of those anchors. When it´s used in the muti-core environment, the advantages of muti-core are not embodied, and the speed isn´t as expected. Mining the potential parallelism of MUMmer algorithm, the paper puts forward a method of parallelism to make it suitable for muti-core architecture, thus to get results as quickly as possible. Experiments show that compared with serial algorithm, the proposed parallel algorithm has been greatly improved in terms of speed, and is more suitable for muti-core architecture.