Coordinated analysis of exon and intron data reveals novel differential gene expression changes

Hamid R Eghbalnia,Karol Mackey,Piotr Chomczynski,William W Wilfinger

doi:10.1038/s41598-020-72482-w

Hamid R Eghbalnia, Karol Mackey + Show 2 more

Open Access

https://doi.org/10.1038/s41598-020-72482-w

Copy DOI

Abstract

RNA-Seq expression analysis currently relies primarily upon exon expression data. The recognized role of introns during translation, and the presence of substantial RNA-Seq counts attributable to introns, provide the rationale for the simultaneous consideration of both exon and intron data. We describe here a method for the coordinated analysis of exon and intron data by investigating their relationship within individual genes and across samples, while taking into account changes in both variability and expression level. This coordinated analysis of exon and intron data offers strong evidence for significant differences that distinguish the profiles of the exon-only expression data from the combined exon and intron data. One advantage of our proposed method, called matched change characterization for exons and introns (MEI), is its straightforward applicability to existing archived data using small modifications to standard RNA-Seq pipelines. Using MEI, we demonstrate that when data are examined for changes in variability across control and case conditions, novel differential changes can be detected. Notably, when MEI criteria were employed in the analysis of an archived data set involving polyarthritic subjects, the number of differentially expressed genes was expanded by sevenfold. More importantly, the observed changes in exon and intron variability with statistically significant false discovery rates could be traced to specific immune pathway gene networks. The application of MEI analysis provides a strategy for incorporating the significance of exon and intron variability and further developing the role of using both exons and intron sequencing counts in studies of gene regulatory processes.

Highlights

RNA sequencing (RNA-Seq) expression analysis currently relies primarily upon exon expression data
We demonstrate that when exons counts are normalized by the corresponding intron counts, the resulting co-expression profile ( log[2] scale) in a control population is markedly different from the profile of exon-only counts
These results demonstrate that relationships between exon and intron counts, and the variability of counts as measure by GINI should be considered to be parameters of interest

Summary

Introduction

RNA-Seq expression analysis currently relies primarily upon exon expression data. The recognized role of introns during translation, and the presence of substantial RNA-Seq counts attributable to introns, provide the rationale for the simultaneous consideration of both exon and intron data. We describe here a method for the coordinated analysis of exon and intron data by investigating their relationship within individual genes and across samples, while taking into account changes in both variability and expression level. The application of MEI analysis provides a strategy for incorporating the significance of exon and intron variability and further developing the role of using both exons and intron sequencing counts in studies of gene regulatory processes. Incorporation of intronic read data has been used to determine RNA Velocity—the time derivative of gene e xpression[31] These recent studies, performed in a variety of organisms, have delineated a new model in which for a subset of genes the most important regulatory sequences are located within introns. Using poly-adenylated RNA for analysis can improve signals from mature RNA, but this may come at the expense of missing functionally important non-coding RNA fragments present in the nucleus and cytoplasm

Methods

Results

Conclusion