摘要
正则化是scRNA-seq数据分析的核心并影响决定下游分析的质量.相比bulkRNA-seq,由于scRNA-seq的zero inflation,其正则化是一个尚未解决的问题.本研究给出了一个bias analysis framework对现有的scRNA-seq正则化方法进行评估比较.这个bias analysis framework对scRNA-seq正则化提供了理论基础.同时作者比较了广为使用的bulkRNA-seq正则化方法,以及专为scRNA-seq设计的正则化方法在scRNA-seq基准数据聚类中的作用.
Normalization is an essential component for single-cell RNA-seq (scRNA-seq) analysis and determines the the quality of downstream analysis. However, scRNA-seq normalization remains a challenge compared to bulk RNA-seq for its zero inflation. In this study, authors review the existing normalization methods for scRNA-seq data under a framework of bias analysis. The framework categorizes different biases in scRNA-seq and provides a solid theoretical foundation for scRNA-seq normalization. This study also compares well-established bulk RNA-seq normalization procedures and tailored scRNA-seq normalization methods on benchmark scRNA-seq data on behalf of phenotype clustering.
作者
Han Henry
刘文斌
LIU Wen-bin(Department of Computer and Information Science,Fordham University,New York 10023,USA;Institute of Computational Science and Technology,Guangzhou University,Guangzhou 510006,China)
出处
《广州大学学报（自然科学版）》
CAS
2019年第2期44-55,共12页
Journal of Guangzhou University:Natural Science Edition
