Variable-Order de Bruijn Graphs

Variable-Order de Bruijn Graphs

Today’s must read algorithm paper ! It describes the theory behind -MEGAHIT: An Ultra-fast Single-node Solution for Large and Complex Metagenomics Assembly via Succinct de Bruijn Graph . [Correction: The algorithm is an improvement over MEGAHIT. Please see comment section.]

Our Contribution. SPAdes [2] and IDBA [21] represent the state-of-the- art for genome assemblers, producing assemblies of greatly improved quality compared to previous approaches. However, their need to construct several de Bruijn graphs of different orders over the assembly process makes them extremely slow on large genomes. In this paper we address this problem by describing a succinct data structure that, for a given K, eciently represents all the de Bruijn graphs for k K and allows navigation within and between each graph. We have implemented our data structure and show that in practice it requires around twice the space of a graph for a single K, and incurs a modest slow down in construction time and on navigation operations.



The de Bruijn graph GK of a set of strings S is a key data structure in genome assembly that represents overlaps between all the K-length substrings of S. Construction and navigation of the graph is a space and time bottleneck in practice and the main hurdle for assembling large, eukaryote genomes. This problem is compounded by the fact that state-of-the-art assemblers do not build the de Bruijn graph for a single order (value of K) but for multiple values of K. More precisely, they build d de Bruijn graphs, each with a specific order, i.e., GK1,GK2,,GKd. Although, this paradigm increases the quality of the assembly produced, it increases the memory by a factor of d in most cases. In this paper, we show how to augment a succinct de Bruijn graph representation by Bowe et al. (Proc. WABI, 2012) to support new operations that let us change order on the fly, effectively representing all de Bruijn graphs of order up to some maximum K in a single data structure. Our experiments show our variable-order de Bruijn graph only modestly increases space usage, construction time, and navigation time compared to a single order graph.

Written by M. //