基于CUDA的生物序列数据算术编码并行压缩

发布时间：2018-05-28 03:37

本文选题：算术编码 + 生物序列　；参考：《计算机应用与软件》2016年12期

【摘要】：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。
[Abstract]:With the development of next generation biological sequence sequencing technology, large-file biological sequence data are becoming more and more common. Although the compression of sequence data can reduce the data storage space, it is difficult for the traditional data compression method to complete the large-scale data compression quickly. Therefore, how to shorten the time of data compression is an important research direction of the current compression technology. The arithmetic coding is realized by using CUDA technique, the characteristics of nucleotide biological sequence data are analyzed, the distribution probability of nucleotides in biological sequence data sets of different species and database is given, and three parallel compression methods are proposed and compared. It is pointed out that the parallel compression method with prior probability has better compression performance. The experimental results show that the parallel compression method based on priori probability not only has high time efficiency, but also can maintain a high data compression rate, and it can solve the problem of fast and efficient compression of large scale biological sequence files.
【作者单位】：上海大学计算机工程与科学学院;上海众恒信息产业股份有限公司;
【基金】：国家自然科学基金重大研究计划项目(91330116) 教育部留学回国人员科研启动基金
【分类号】：Q811.4;TP301.6

【相似文献】