De Bruijn graph construction is a basic component in de novo genome assembly for short reads generated from the second-generation sequencing machines. As this component processes a large amount of data and performs intensive computation, we propose to use the GPU (Graphics Processing Unit) for acceleration. Specifically, we propose a staged algorithm to utilize the GPU for computation over large data sets that do not fit into the GPU memory. We also pipeline the I/O, GPU, and CPU processing to further improve the overall performance. Our preliminary results show that our GPU-accelerated graph construction on an NVIDIA S1070 server achieves a speedup of around two times over previous performance results on a 1024-node IBM Blue Gene/L. © 2013 Springer-Verlag.
CITATION STYLE
Lu, M., Luo, Q., Wang, B., Wu, J., & Zhao, J. (2013). GPU-accelerated bidirected De Bruijn graph construction for genome assembly. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7808 LNCS, pp. 51–62). https://doi.org/10.1007/978-3-642-37401-2_8
Mendeley helps you to discover research relevant for your work.