Building approximate overlap graphs for DNA assembly using random-permutations-based search

between two reads indicates an approximate overlap between the reads. Since the algorithm finds approximate overlaps directly, it can process reads without errorcorrection preprocessing steps. Extensions of the algorithm, such as construction graphs of overlapping pairs of reads, are discussed. The algorithm can be used to construct graphs for assembly and for other related applications such as error correction. Preliminary experimental results indicate that the algorithm constructs graphs which lead to relatively long contigs even in the presence of sequencing errors.