Efficient algorithm for circular burrows-wheeler transform

Abstract

Given a set of d patterns, the circular dictionary matching problem is to index such that for any online query text T, we can quickly locate the occurrences of any cyclic shift of any pattern of within T efficiently. This problem can be applied on practical problems that arise in bioinformatics and computational geometry. Recently, Hon et al. (2011) applied a variant of the well-known Burrows-Wheeler transform, called circular Burrows-Wheeler transform (circular BWT) [Mantaci, Restivo, Rosone, and Sciortino, Theoretical Computer Science, 2007], and showed that this can be used to solve the circular dictionary matching problem efficiently. In this paper, we give the first construction algorithm for the circular BWT, which takes O(nlogn) time and requires O(nlogσ) bits working space, where n denotes the total length of the patterns in and σ is the alphabet size. © 2012 Springer-Verlag.