基于CUDA的生物序列数据算术编码并行压缩

摘要

With the development of the next generation of bio-sequencing technology,files of biological sequence data become larger and larger.The compression of sequence data can reduce data storage space,however,it is difficult for traditional method to complete compression quickly.It is an important direction of the current compression technology research to shorten the processing time of data compression.Thus,the parallel arithmetic coding compression is realized by using CUDA and combining characteristics of biological sequence data.Three parallel compression methods are compared and it is found that the method based on priority probabilities can achieve rapid compression of large-scale biological sequence data,in which the priori probabilities depend on the type of sequences such as species and databases.Experiments demonstrate that the compression of GPU parallel arithmetic encoding method based on priori probabilities not only has the greater time efficiency,but also has high compression ratio,which can address the problem of compressing large biological sequence files rapidly and efficiently.%随着下一代生物序列测序技术的发展，大文件生物序列数据越来越常见。虽然压缩序列数据能减少数据存储空间，但是传统的数据压缩的方法很难快速完成大规模的序列压缩，因此如何缩短数据压缩时间是当前压缩技术研究的一个重要方向。采用CUDA 技术实现算术编码，分析核苷酸生物序列数据特性，给出不同物种及数据库生物序列数据集中核苷酸的分布概率，提出并比较三种并行压缩方法，指出先验概率的并行压缩方法具有更好的压缩性能。实验结果表明，先验概率的并行压缩方法不仅具有较高的时间效率，而且也能保持较高的数据压缩率，能较好地解决大规模生物序列文件的高效快速压缩问题。