PUB - Publikationen an der Universität Bielefeld

Motivation: The increasing amount of next-generation sequencing data poses a fundamental challenge on large scale genomic analytics. Existing tools use different distributed computational platforms to scale-out bioinformatics workloads. However, the scalability of these tools is not efficient. Moreover, they have heavy run time overheads when pre-processing large amounts of data. To address these limitations, we have developed Sparkhit: a distributed bioinformatics framework built on top of the Apache Spark platform.
Results: Sparkhit integrates a variety of analytical methods. It is implemented in the Spark extended MapReduce model. It runs 92 to 157 times faster than MetaSpark on metagenomic fragment recruitment and 18 to 32 times faster than Crossbow on data pre-processing. We evaluated sensitivity and accuracy with the simulated data provided here.