Efficient bubble enumeration in directed graphs

Polymorphisms in DNA- or RNA-seq data lead to recognisable patterns in a de Bruijn graph representation of the reads obtained by sequencing. Such patterns have been called mouths, or bubbles in the literature. They correspond to two vertex-disjoint directed paths between a source s and a target t. Due to the high number of such bubbles that may be present in real data, their enumeration is a major issue concerning the efficiency of dedicated algorithms. We propose in this paper the first linear delay algorithm to enumerate all bubbles with a given source.