Advantages of distributed and parallel algorithms that leverage Cloud Computing platforms for large-scale genome assembly.

Background: The transition to Next Generation sequencing (NGS) sequencing technologies has had numerous applications in Plant, Microbial and Human genomics during the past decade. However, NGS sequencing trades high read throughput for shorter read length, increasing the difficulty for genome assembly. This research presents a comparison of traditional versus Cloud computing-based genome assembly software, using as examples the Velvet and Contrail assemblers and reads from the genome sequence of the zebrafish ( Danio rerio ) model organism. Results: The first phase of the analysis involved a subset of the zebrafish data set (2X coverage) and best results were obtained using K-mer size of 65, while it was observed that Velvet takes less time than Contrail to complete the assembly. In the next phase, genome assembly was attempted using the full dataset of read coverage 192x and while Velvet failed to complete on a 256GB memory compute server, Contrail completed but required 240hours of computation. Conclusion: This research concludes that for deciding on which assembler software to use, the size of the dataset and available computing hardware should be taken into consideration. For a relatively small sequencing dataset, such as microbial or small eukaryotic genome, the Velvet assembler is a good option. However, for larger datasets Velvet requires large-memory  compute servers in the order of 1000GB or more. On the other hand, Contrail is implemented  using Hadoop, which performs the assembly in parallel across nodes of a compute cluster. Furthermore, Hadoop clusters can be rented on-demand from Cloud computing providers, and therefore Contrail can provide a simple and cost effective way for genome assembly of data generated at laboratories that lack the infrastructure or funds to build their own clusters.