When we find a new DNA sequence, the most important thing is to find its function. To predict the function of the new sequence, we compare it with the database sequences whose functions are known.
At the present, the amount of DNA sequences has been increasing exponentially. Even though BLAST, a sequence search tool, is impressively good so far, it would reach its limit under the current condition before long.
The method of high speed DNA database search using filtering, which is presented in the thesis, would be a good way to overcome this condition.
A small proportion of database sequences selected by filtering, which are possibly homologous with the query sequence, are passed to the sequence search tool. So the search tool can complete its work quickly, because the sequence search tool searches actually a small proportion of database sequences. The information for filtering can be made by preprocessing.
To ensure the performance of our new filtering method, it was applied to BLAST which is the most popular searching tool. The new filtering method could be complementary for the current sequence search tools.
