Bounds for resequencing by hybridization

Abstract

We study the problem of finding the sequence of an unknown DNA fragment given the set of its k-long subsequences and a homologous sequence, namely a sequence that is similar to the target sequence. Such a sequence is available in some applications, e.g., when detecting single nucleotide polymorphisms. Pe'er and Shamir studied this problem and presented a heuristic algorithm for it. In this paper, we give an algorithm with provable performance: We show that under some assumptions, the algorithm can reconstruct a random sequence of length O(4k) with high probability. We also show that no algorithm can reconstruct sequences of length Ω(log k· 4k). © Springer-Verlag Berlin Heidelberg 2003.