We propose a formal model and an algorithm for detecting inversion breakpoints without a reference genome, directly from raw NGS data. This model is characterized by a fixed size topological pattern in the de Bruijn Graph. We describe precisely the possible sources of false pos- itives and false negatives and we additionally propose a sequence-based filter giving a good trade-off between precision and recall of the method. We implemented these ideas in a prototype called TakeABreak. Ap- plied on simulated inversions in genomes of various complexity (from E. coli to a human chromosome dataset), TakeABreak provided promising results with a low memory footprint and a small computational time.