A new view of transcriptome complexity and regulation through the lens of local splicing variations

Abstract

Alternative splicing (AS) can critically affect gene function and disease, yet mapping splicing variations remains a challenge. Here, we propose a new approach to define and quantify mRNA splicing in units of local splicing variations (LSVs). LSVs capture previously defined types of alternative splicing as well as more complex transcript variations. Building the first genome wide map of LSVs from twelve mouse tissues, we find complex LSVs constitute over 30% of tissue dependent transcript variations and affect specific protein families. We show the prevalence of complex LSVs is conserved in humans and identify hundreds of LSVs that are specific to brain subregions or altered in Alzheimer's patients. Amongst those are novel isoforms in the Camk2 family and a novel poison exon in Ptbp1, a key splice factor in neurogenesis. We anticipate the approach presented here will advance the ability to relate tissue-specific splice variation to genetic variation, phenotype, and disease.DOI: http://dx.doi.org/10.7554/eLife.11752.001

Ancient antagonism between CELF and RBFOX families tunes mRNA splicing outcomes

“…Therefore, we conclude that CELF2 has a widespread and consistent impact on alternative splicing in both unstimulated and stimulated T cells. For subsequent analysis, we have thus merged together the high-confidence LSVs from unstimulated and stimulated cells (i.e., those with a probability of >95% that the difference in inclusion (delta PSI or ΔΨ) is >20%) to identify sets of cassette exons for which inclusion is increased (CELF2-repressed) or decreased (CELF2-enhanced) upon depletion of (Vaquero-Garcia et al 2016) that showed significantly altered splicing (|ΔΨ| ≥ 20% with probability ≥95%) upon depletion of CELF2 by shRNA in unstimulated (left), stimulated (right), or both (middle) JSL1 Jurkat T cells (Fisher's exact (FE) test, two-tailed P < 7 × 10 −192 ). (B) Scatterplot comparing E(ΔΨ) values for the most changing junction from the 271 unique, coregulated LSVs upon CELF2 depletion in unstimulated (X) and stimulated (Y ) Jurkat T cells.…”

“…This resulted in 265 CELF2-repressed, 371 CELF2-enhanced, and 1668 CELF2-unresponsive exons (Supplemental Table S1; see Methods). We note that the final number of CELF2-regulated events defined this way (265 + 371 = 636) is less than those in Figure 1A due to restricting the list to cassette exons (to facilitate subsequent analysis) and to the fact that multiple LSVs can report on the same cassette exon (Vaquero-Garcia et al 2016).…”

Whippet: an efficient method for the detection and quantification of alternative splicing reveals extensive transcriptomic complexity

“…Our results from applying Whippet indicate that high-entropy AS events occur more frequently in vertebrate transcriptomes than previously appreciated 12,31 , and further provide evidence that these events are likely biologically significant since their entropy levels are frequently tissue-regulated, conserved, and the corresponding variant transcripts are highly expressed. Many of the events are reminiscent of wellstudied examples of high-entropy AS in other systems, such as the myriad of splice variants generated by tandem arrays of alternative exons in the Drosophila DSCAM gene 32 .…”

“…3a). To assess the accuracy of AS-quantification, we compared published RT-PCR-derived Y values for AS events in liver and cerebellum samples with Whippet-derived Y values obtained from RNAseq data from the same tissue types 12 . Whippet-derived Y values correlated highly with the RT-PCR measurements (r=0.963, Pearson's correlation coefficient) and with a comparable error rate as the best other methods tested (Fig.…”

“…4b, p > 0.05, Wilcoxon rank-sum one sided test, Whippet vs other algorithms). To this end, we compared its speed and memory usage to those of eight other highly-cited splicing tools 9,12,[15][16][17][18][19] when analyzing several paired-end RNA-seq datasets from HeLa cells with increasing read depth (~10M, ~25M and ~50M). Remarkably, Whippet quantifies AS on an eventspecific basis from a raw paired-end 25M RNA-seq read dataset in less than forty-five minutes on a typical cluster node (Dual-Core AMD Opteron(tm) Processor 8218, 2.5…”