On using two-phase filtering in indexed approximate string matching with application to searching unique oligonucleotides

We discuss using an indexing scheme to accelerate approximate search over a static text in the case of using unit cost edit distance as the measure of similarity between strings. First we generally consider the filtering criteria that can be used as a basis for the index, and then propose using filtering twice before the final checking phase. The last part consists of presenting an indexed approximate string matching application in bioinformatics, which is the search of unique oligonucleotides. We present practical comparisons and results for using different filtering schemes in this application. Our tests have involved a total of 15 different genomes, from which we present some results involving the largest two of these: The genome of Saccharomyces cerevisiae (baker's yeast) and a recent draft of the human genome, the latter being also the main target of the application.