RNA-seq has become the most prevalent technology for measuring genome-wide gene expression, but the best practices for processing and analysing RNA-seq data are still an open question. Many statistical methods have been developed to identify genes differentially expressed between sample groups from RNA-seq data. These methods differ by their data distribution assumptions, choice of statistical test, and computational resource requirements. Over 25 methods of differential expression detection were validated and made available through a user-friendly web portal, RNA-seq 2G. All methods are suitable for analysing differential gene expression between two groups of samples. They commonly use a read count matrix derived from RNA-seq data as input and statistically compare groups for each gene. The web portal uses a Shiny app front-end and is hosted by a cloud-based server provided by Amazon Web Service. The comparison of methods showed that the data distribution assumption is the major determinant of differences between methods. Most methods are more likely to find that longer genes are differentially expressed, which substantially impacts downstream gene set-level analysis. Combining results from multiple methods can potentially diminish this bias. RNA-seq 2G makes the analysis of RNA-seq data more accessible and efficient, and is freely available at http://rnaseq2g.awsomics.org.

RNA-Seq 2G: Online Analysis Of Differential Gene Expression With Comprehensive Options Of Statistical Methods