Evaluation of tools for highly variable gene discovery from single-cell RNA-seq data

Abstract Traditional RNA sequencing (RNA-seq) allows the detection of gene expression variations between two or more cell populations through differentially expressed gene (DEG) analysis. However, genes that contribute to cell-to-cell differences are not discoverable with RNA-seq because RNA-seq sam...

Full description

Saved in:
Bibliographic Details
Published in:Briefings in bioinformatics Vol. 20; no. 4; pp. 1583 - 1589
Main Authors: Yip, Shun H, Sham, Pak Chung, Wang, Junwen
Format: Journal Article
Language:English
Published: England Oxford University Press 19.07.2019
Oxford Publishing Limited (England)
Subjects:
Animals
Cluster Analysis
Computational Biology
Computer Simulation
Databases, Nucleic Acid - statistics & numerical data
Embryo cells
Gene expression
Gene Expression Profiling - statistics & numerical data
Gene sequencing
Genes
Genetic diversity
Genetic Variation
Humans
Quorum sensing
Reproducibility of Results
Ribonucleic acid
RNA
RNA-Seq - statistics & numerical data
Single-Cell Analysis - statistics & numerical data
Software
Software packages
Software Review
Stem cells
highly variable gene
scRNA-seq
single-cell RNA seq
software
DEG analysis
ISSN:1467-5463, 1477-4054, 1477-4054
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Traditional RNA sequencing (RNA-seq) allows the detection of gene expression variations between two or more cell populations through differentially expressed gene (DEG) analysis. However, genes that contribute to cell-to-cell differences are not discoverable with RNA-seq because RNA-seq samples are obtained from a mixture of cells. Single-cell RNA-seq (scRNA-seq) allows the detection of gene expression in each cell. With scRNA-seq, highly variable gene (HVG) discovery allows the detection of genes that contribute strongly to cell-to-cell variation within a homogeneous cell population, such as a population of embryonic stem cells. This analysis is implemented in many software packages. In this study, we compare seven HVG methods from six software packages, including BASiCS, Brennecke, scLVM, scran, scVEGs and Seurat. Our results demonstrate that reproducibility in HVG analysis requires a larger sample size than DEG analysis. Discrepancies between methods and potential issues in these tools are discussed and recommendations are made.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ObjectType-Review-3
content type line 23
ISSN:1467-5463
1477-4054
1477-4054
DOI:10.1093/bib/bby011