Pathset Graphs: A Novel Approach for Comprehensive Utilization of Paired Reads in Genome Assembly

Abstract

One of the key advances in genome assembly that has led to a significant improvement in contig lengths has been improved algorithms for utilization of paired reads (mate-pairs). While in most assemblers, mate-pair information is used in a post-processing step, the recently proposed Paired de Bruijn Graph (PDBG) approach incorporates the mate-pair information directly in the assembly graph structure. However, the PDBG approach faces difficulties when the variation in the insert sizes is high. To address this problem, we first transform mate-pairs into edge-pair histograms that allow one to better estimate the distance between edges in the assembly graph that represent regions linked by multiple mate-pairs. Further, we combine the ideas of mate-pair transformation and PDBGs to construct new data structures for genome assembly: pathsets and pathset graphs.

Overlap graphs and de Bruijn graphs: data structures for de novogenome assembly in the big data era

“…Many different methods were proposed in the literature to deal with repeated regions of the genome and, surely, many more will be proposed in the next years. For example, IDBA [31] solves this problem by iterating the construction and analysis of the de Bruijn graph for increasing values of k , Paired de Bruijn Graphs [32] includes mate pair information in the structure of the graph, Pathset graphs [33] exploit the mate pair information to estimate the distance between edges in the assembly graph, SEQuel [34] almost completely avoids this problem by using an extension of the de Bruijn graph (called positional de Bruijn Graph) that incorporate the approximate positions of the reads in the sequenced genome, and AlignGraph [35] distinguishes paths for repetitive regions by incorporating mate pair information and alignment position in the nodes of the graph.…”