Retained introns in long RNA-seq reads are not reliably detected in sample-matched short reads

Abstract

Background There is growing interest in retained introns in a variety of disease contexts including cancer and aging. Many software tools have been developed to detect retained introns from short RNA-seq reads, but reliable detection is complicated by overlapping genes and transcripts as well as the presence of unprocessed or partially processed RNAs. Results We compared introns detected by 8 tools using short RNA-seq reads with introns observed in long RNA-seq reads from the same biological specimens. We found significant disagreement among tools (Fleiss’ $$\kappa = 0.113$$ κ = 0.113 ) such that 47.7% of all detected intron retentions were not called by more than one tool. We also observed poor performance of all tools, with none achieving an F1-score greater than 0.26, and qualitatively different behaviors between general-purpose alternative splicing detection tools and tools confined to retained intron detection. Conclusions Short-read tools detect intron retention with poor recall and precision, calling into question the completeness and validity of a large percentage of putatively retained introns called by commonly used methods.

Single-cell long-read sequencing in human cerebral organoids uncovers cell-type-specific and autism-associated exons

“… 43 However, it has been challenging to quantify introns by short-read sequencing. 44 Using the long-read sequencing data of human cerebral organoids, we identified IRs by IRFinder-S 45 and compared IRs from different cell types. A total of 1,427 differentially retained introns (DRIs) were identified (∣ΔPIR∣ ≥ 5% and FDR < 0.05; Figure 3A ; Table S4 ).…”

“…It is challenging to study IR using short-read sequencing due to repetitive sequences. 44 In principle, long-read sequencing covers the full transcript and provides a promising solution. Our analysis of human cerebral organoids uncovered over 1,400 differential IR events in oligoT-captured transcripts, suggesting the robustness.…”

Profiling genetically driven alternative splicing across the Indonesian Archipelago

“…Our detection of alternative splicing revealed that the most frequent splicing event types were skipped exon events (50.1%) and the least frequent were retained intron events (3.5%). While these proportions are expected in higher eukaryotes [41, 118–123], it is important to note that short-read algorithms for detecting AS events are less likely to misassemble skipped exon events and can therefore identify them with high accuracy, but they suffer from poor precision with respect to RI detection [124, 125]. Accurately disambiguating real splicing events from background noise (such as partial transcript processing) is greatly facilitated by long-read sequencing technologies, which enable much more precise discovery and reconstruction of full-length isoforms, significantly improving the characterization of functional transcripts [126].…”

Global Analysis of Dark- and Heat-Regulated Alternative Splicing in Arabidopsis

“…IR was also found to be the most abundant DAS events during heat stress in our study, though we did not observe a drastic increase in the number of IR, probably reflecting the limitation of our analysis. Moreover, it has been recently reported that IRs are not reliably detected by short RNA-sequencing reads compared to long RNA-sequencing reads [ 104 ]. In the future, to detect more novel transcripts, a combined assembly strategy that merges a reference-based assembly and de novo transcriptome assembly can also be applied to create a more comprehensive transcriptome for AS analysis, especially under specific stress conditions [ 105 ].…”