iREAD: a tool for intron retention detection from RNA-seq data

Abstract

Background: Intron retention (IR) has been traditionally overlooked as 'noise' and received negligible attention in the field of gene expression analysis. In recent years, IR has become an emerging field for interrogating transcriptomes because it has been recognized to carry out important biological functions such as gene expression regulation and it has been found to be associated with complex diseases such as cancers. However, methods for detecting IR today are limited. Thus, there is a need to develop novel methods to improve IR detection. Results: Here we present iREAD (intron REtention Analysis and Detector), a tool to detect IR events genome-wide from high-throughput RNA-seq data. The command line interface for iREAD is implemented in Python. iREAD takes as input a BAM file, representing the transcriptome, and a text file containing the intron coordinates of a genome. It then 1) counts all reads that overlap intron regions, 2) detects IR events by analyzing the features of reads such as depth and distribution patterns, and 3) outputs a list of retained introns into a tab-delimited text file. iREAD provides significant added value in detecting IR compared with output from IRFinder with a higher AUC on all datasets tested. Both methods showed low false positive rates and high false negative rates in different regimes, indicating that use together is generally beneficial. The output from iREAD can be directly used for further exploratory analysis such as differential intron expression and functional enrichment. The software is freely available at https://github.com/genemine/iread. Conclusion: Being complementary to existing tools, iREAD provides a new and generic tool to interrogate poly-A enriched transcriptomic data of intron regions. Intron retention analysis provides a complementary approach for understanding transcriptome.

Intron Retention as a Mode for RNA-Seq Data Analysis

“…Currently, tools dedicated to IR detection are available (Bai et al, 2015;Pimentel et al, 2015a;Middleton et al, 2017;Li et al, 2020). Bai et al (2015) developed IRcall (a ranking strategy) and IRclassifier (random forest classifiers) to detect IR events.…”

“…It can combine biological replicates to reduce the number of false positives. The isoform quantification and analysis of IR are performed in different software environments, which may be inconvenient (Li et al, 2020). IRFinder (Middleton et al, 2017) provides a complete pipeline for identifying IR events, including genome preparation, data preparation and quality control, IR quantification, and differential analysis.…”

“…IRFinder identifies IR candidates based on IR ratio and the number of intronic reads. iREAD (Li et al, 2020) uses the Shannon entropy to quantify the uniformity of the distribution of reads across the intron. To avoid ambiguity, only the independent intron that does not overlap with any exon of any gene is considered in iREAD.…”

“…Currently, the detection of IR is based on computational analysis of high-throughput RNA-seq data. In recent years, tools dedicated to IR detection have been developed, such as IRCall and IR classifier (Bai et al, 2015), Keep Me Around (KMA) (Pimentel et al, 2015a), intron Retention Analysis and Detector (iREAD) (Li et al, 2020), and IRFinder (Middleton et al, 2017). In addition, some tools originally designed to detect AS events can also be used to detect IR, such as mixture-of-isoforms (MISO) (Katz et al, 2010), multivariate analysis of transcript splicing (MATS) (Shen et al, 2012), replicate MATS (rMATS) (Shen et al, 2014), comprehensive alternative splicing hunting (CASH) (Wu et al, 2017), and DEXSeq (Anders et al, 2012).…”

Intron retention and its impact on gene expression and protein diversity: A review and a practical guide

“…Because IR detection requires the aforementioned corrections and specific measurements, we highly recommend the use of IR dedicated software over more general splicing detection algorithms. Table 2 lists a subset of IR dedicated software and summarizes their general approach and metric for estimating IR levels (H.-D. Li et al, 2020;Middleton et al, 2017;Oghabian et al, 2018;Pimentel et al, 2015).…”