Research On Biological Sequential Pattern Mining Algorithms And Their Applications

With the development of Bioinformatics, the data of biological science is exploding, which force people to search more powerful tools for administrating and analyzing. Data Mining is the most effective way to analyze data, and it can be used to search all kinds of knowledge behind the data. In the analysis of biological sequences, Data Mining is used widely. It can improve the ability of dealing data, and play an important role on producing valuable biological knowledge. Biological Sequential Pattern Mining is one of the important research fields in Biological Sequential Data Mining, it is important for identifying genes, element in no coding field and information in protein sequences.Both Mining Frequent Patterns and Searching Tandem Repeats are import research fields in Biological Sequential Pattern Mining. Traditional Ming Frequent Patterns algorithms will generate lots of patterns with short length in the process of mining which cause the low efficiency of mining. In order to overcome the lack of traditional algorithms, the algorithm MBioPM was proposed. We used a method called "Motif-divide" to start mining patterns with given length, which avoided producing lots of patterns with short length. The theoretical analysis and experimental results show that MBioPM improves the performance.Traditional Searching Tandem Repeats algorithms need to compare patterns on all of them which affect the performance of mining. To attack the problem of traditional algorithms, we propose the algorithm MSATR. In the process of MSATR, Tandem Repeats can be mined just with comparing the patterns next to each other, and because of that, MSATR improves the efficiency when compared to traditional algorithms. The performance of MSATR was proved by the theoretical analysis and experiment.