Using cascading Bloom filters to improve the memory usage for de Brujin graphs

Kamil Salikhov (1) , Gustavo Sacomoto (2, 3) , Gregory Kucherov (4, 5)

1 MSU - Lomonosov Moscow State University
2 Baobab
3 BAMBOO - An algorithmic view on genomes, cells, and environments
4 Department of Computer Science [Beer-Sheva]
5 LIGM - Laboratoire d'Informatique Gaspard-Monge

Background
De Brujin graphs are widely used in bioinformatics for processing next-generation sequencing data. Due to a very large size of NGS datasets, it is essential to represent de Bruijn graphs compactly, and several approaches to this problem have been proposed recently.
Results
In this work, we show how to reduce the memory required by the data structure of Chikhi and Rizk (WABI'12) that represents de Brujin graphs using Bloom filters. Our method requires 30% to 40% less memory with respect to their method, with insignificant impact on construction time. At the same time, our experiments showed a better query time compared to the method of Chikhi and Rizk.
Conclusion
The proposed data structure constitutes, to our knowledge, currently the most efficient practical representation of de Bruijn graphs.

Domaines: Biologie moléculaire

Kamil Salikhov, Gustavo Sacomoto, Gregory Kucherov. Using cascading Bloom filters to improve the memory usage for de Brujin graphs. Algorithms for Molecular Biology, 2014, 9 (1), pp.2. ⟨10.1186/1748-7188-9-2⟩. ⟨hal-00971576⟩