深度学习在DNA存储读段重建的应用
The application of deep learning in reads reconstruction for DNA storage
姚翔宇 1刘希晨 2昝乡镇 1许鹏 1刘文斌1
作者信息
- 1. 广州大学计算科技研究院,广东广州 510006
- 2. 广州商学院信息技术与工程学院,广东广州 511363
- 折叠
摘要
DNA存储技术是一项着眼于未来的具有划时代意义存储技术,它将数字信息编码为核苷酸序列,然后通过化学合成将序列写入DNA分子,最后通过DNA测序技术读取信息.相较于电子存储技术,其在信息密度、数据安全性以及保存年限等方面具有极大的优势.然而在DNA存储技术的数据读取端,测序读段存在着大量碱基替换、插入和删除错误,因此需进行读段重建来恢复原始数据.读段重建方法要求高成功率以及一定的时效性,以实现文件可靠存取并提高DNA存储技术的读写效率.文章介绍了现有的基于深度学习的读段重建模型,通过对模型架构、重建理念以及纠错能力等方面的比较指出了目前研究的局限性,并展望了未来深度学习在读段重建中可能的研究方向.
Abstract
DNA storage technology is a promising new type of data storage technology,it encodes dig-ital information into nucleotide sequences,then writes the sequences into DNA molecules through chemical synthesis,and finally reads the information through DNA sequencing technology.Compared to electronic storage technology,it has great advantages in information density,data security,and li-fespan.However,it outputs a large number of sequencing reads with base errors(substitution,inser-tion and deletion),therefore,to restore the original data from erroneous reads,reads reconstruction is usually performed.Accurate read reconstruction requires high success rate and time efficiency to a-chieve reliable file access and accelerate the reading and writing process of DNA storage technology.This paper introduces the existing deep learning-based reads reconstruction models.By comparing ar-chitecture,basic concepts and error correction abilities,we point out the limitations of these methods and discuss the prospects for future research directions.
关键词
DNA存储技术/碱基错误/读段重建/深度学习Key words
DNA storage/base error/reads reconstruction/deep learning引用本文复制引用
基金项目
国家自然科学基金(62072128)
国家自然科学基金(62002079)
国家自然科学基金(62102104)
出版年
2024
NETL
NSTL
万方数据