Figure 1

Fingerprint computation on b -length substrings. When looking for overlaps between S and r, the fingerprints are computed on the substrings r[x, . . ., x+b - 1] and S[y, . . ., y + b - 1], respectively, where x and b are set before the contig's extension phase. We require an (almost) exact b-length match between r and S in order to include r in the set of putative overlapping reads, by setting f _H (r[x, . . ., x + b - 1]) = f _H (S[y, . . ., y + b -1]). Using such a method, the suffix-prefix overlaps that can be detected are those of length l ≥ x + b.