Graph-based models of the Oenothera mitochondrial genome capture the enormous complexity of higher plant mitochondrial DNA organization

Plant mitochondrial genomes display an enormous structural complexity, as recombining repeat pairs lead to the generation of various sub-genomic molecules. This makes these genomes extremely challenging to assemble. Here, we present a novel pipeline called SAGBAC (Semi-Automated Graph-Based Assembly Curator) that identifies recombinogenic repeat pairs and reconstructs plant mitochondrial genomes. The pipeline processes and filters non-graph-based assembly outputs using our novel ISEIS (Iterative Sequence Ending Identity Search) algorithm to obtain a graph-based visualization. We applied this approach to three mitochondrial genomes of the evening primrose (Oenothera), a model organism for cytoplasmic genetics. We could show that all identified repeat pairs are flanked by two alternative unique sequences (contigs) defining so-called "ouble forks". This leads to four contig-repeat-contig combinations for each repeat pair. Based on the structural model, stoichiometry of the different contig-repeat-contig combinations was analysed using Illumina mate-pair and PacBio RSII data. This uncovered remarkable structural diversity of the closely related mitochondrial genomes, as well as substantial phylogenetic variation of the underlying repeat units. Our model allows predicting all recombination events and, thus, all possible sub-genomes. In future work, this can contribute to the investigation of sub-genome organization and dynamics in different tissues and at various developmental stages.