作者: Yingbo Cui , Xiangke Liao , Xiaoqian Zhu , Bingqiang Wang , Shaoliang Peng
DOI: 10.1007/978-3-319-07581-5_14
关键词:
摘要: Mapping sequenced reads to a reference genome, also known as sequence alignment, is central for analysis. Emerging sequencing technologies such next generation (NGS) lead an explosion of data, which far beyond the process capabilities existing alignment tools. Consequently, becomes bottleneck Intensive computing power required address this challenge. A key feature that different are independent. Considering property, we proposed multi-level parallelization strategy speed up BWA, widely used tool and developed our massively parallel aligner: mBWA. mBWA contains two levels parallelization: firstly, data input/output (IO) by three-stage pipeline; secondly, enabled Intel Many Integrated Core (MIC) coprocessor technology. In paper, demonstrate outperforms BWA combination those techniques. To best knowledge, first run on MIC it can achieve more than 5-fold speedup over original while maintaining precision.