Sequence analysis Data-dependent bucketing improves reference-free compression of sequencing reads

Motivation: The storage and transmission of high-throughput sequencing data consumes signifi-cant resources. As our capacity to produce such data continues to increase, this burden will only grow. One approach to reduce storage and transmission requirements is to compress this sequencing data. Results: We present a novel technique to boost the compression of sequencing that is based on the concept of bucketing similar reads so that they appear nearby in the file. We demonstrate that, by adopting a data-dependent bucketing scheme and employing a number of encoding ideas, we can achieve substantially better compression ratios than existing de novo sequence compression tools, including other bucketing and reordering schemes. Our method, Mince, achieves up to a 45 % reduction in file sizes (28 % on average) compared with existing state-of-the-art de novo com-pression schemes. Availability and implementation: Mince is written in Cþþ11, is open source and has been made available under the GPLv3 license. It is available a