Querying highly similar sequences

In this paper, we present a solution to the extreme similarity sequencing problem. The extreme similarity sequencing problem consists of finding occurrences of a pattern p in a set S(0), S(1), …, S(k), of sequences of equal length, where S(i), for all 1≤i≤k, differs from S(0) by a constant number of errors - around 10 in practice. We present an asymptotically fast O(n + occ logocc) time algorithm, as well as a practical O(nk/w) time algorithm for solving this problem, where n is the length of a sequence, occ is the number of candidate occurrences reported by our technique, w is the size of the machine word, and the total number of errors is bounded by k - the number of sequences.