Thorough assessment on differential gene expression analysis methods for RNA-seq data

Abstract

Abstract Recently, RNA profiling based on HT sequencing is replacing microarrays for the study of differential gene expression. In practice, millions of ‘reads’ are sequenced from random positions of the input RNAs which are computationally mapped on a reference genome to reveal a ‘transcriptional map’, where the number of reads aligned to each gene gives a measure of its level of expression. The powerful features of RNA-seq, such as high resolution and broad dynamic range, have boosted an unprecedented progress of transcriptomics research, producing an impressive amount of data worldwide. To deal with the different steps of data analysis several computational tools have been developed and updated at a fast pace which is in fact far more complex and consists in several processing steps. In this review, we describe the current RNA-seq analysis framework, focusing on each computational step from read preprocessing to differential expression (DE) analysis. We review the methodologies available, along with their underlying algorithmic strategies and believe this work can provide a broad overview of RNA-seq analysis and can guide users to define and implement their own processing pipeline.