Cloud Computing for De Novo Metagenomic Sequence Assembly

Abstract

In metagenomics, the population sequencing is an approach to recover the genomic sequences in the genetically diverse environment. Combined with the recently developed next generation sequencing platform, mategenomics data analysis has greatly enlarged the size of sequencing datasets and decreased the cost. The complete and accurate assembly of sequenced reads from an environmental sample improves the efficiency of genome functional and taxonomical classification. A common bottleneck of the available tools is the high computing requirement for efficiently assembling vast amounts of data generated from large-scale sequencing projects. To address these limitations, we developed a parallel strategy to accelerate computation and boost accuracy. We also presented an instance of this strategy for a state-of-the-art assembly tool, Genovo, on Apache hadoop platform. As a demonstration of the capability of our approach, we compared the performance of our method to two other short read assembly programs on a series of synthetic and real datasets created using the 454 platform, the largest of which has 683k reads. Under the parallel strategy, the ability of reconstruction of bases outperformed other tools both on speed and several assembly evaluation metrics