Fig. 1

PenguiN overview. A PenguiN proceeds in two stages. During stage I, PenguiN assembles six-frame translated reads (blue). All sequences sharing a k-mer are grouped together. In each k-mer group, all sequences are aligned to the longest one (red), and the best-matching sequences are used to extend the red sequence. Penguin keeps track of the underlying nucleotide sequences, resulting in assembled coding sequences (dark green). During stage II, the coding sequences are combined with nucleotide reads (light green). In both stages, sequences are iteratively extended by computing overlap alignments (black dotted lines) in linear time, ranking possible extensions by their quality, and extending upstream and downstream with the best sequence. B De Bruijn graph assemblers cannot resolve strains across conserved regions longer than their k-mer length (black bar). In contrast, overlap assemblers like PenguiN can link strain assemblies correctly across all conserved regions shorter than the (paired) read length