Analysis of Long Noncoding RNAs in RNA-Seq Data

Abstract

Long noncoding RNAs (lncRNAs) constitute a major fraction of the output of the genome in most higher eukaryotes. However, analysis of their expression in RNA-seq experiments poses unique challenges due to the biological features of this class of RNAs, namely, their generally low expression level, abundance of repeat-element-derived sequences, frequent locus overlap with other transcripts, the presence of a significant non-polyadenylated fraction, and paucity of splicing. While the main computational RNA-seq analysis steps for the study of lncRNAs are similar to those used for the analysis of protein-coding transcriptome, special considerations must be taken at nearly every step to ensure the optimal detection and quantification of lncRNAs. Specifically, changes in sample preparation, library construction, choice of sequencing methodology, and quantification and differential expression algorithms are needed for accurate representation and expression analysis of lncRNA. In addition to the commonly used RNA-seq workflow steps, the study of the lncRNAs involves the lncRNA-specific computational steps of novel gene discovery and analysis of the protein-coding potential of the discovered putative lncRNAs, which are essential for obtaining a complete picture of the noncoding transcriptome. While the traditional RNA-seq will continue to contribute to our understanding of biology of lncRNAs, the increased availability of newer techniques such as capture-seq, single-cell RNA-seq, and in situ RNA-seq is likely to revolutionize our understanding of the regulation and function of this novel and poorly understood class of cellular RNAs.\r