Splice-site Strength Estimation: A simple yet powerful approach to analyse RNA splicing

Abstract

Association Studies.RNA splicing, and variations in this process referred to as alternative splicing, are critical aspects of gene regulation in eukaryotes. From environmental responses in plants to being a primary link between genetic variation and disease in humans, splicing differences confer extensive phenotypic changes across diverse organisms 1-3 . Current approaches for analysing splicing rely on quantifying variant transcripts (i.e., isoforms) or splicing events (i.e., intron retention, exon skipping etc) 4,5 . However, regulation of splicing occurs at the level of selection of individual splice sites, which results in variation in the abundance of isoforms and/or splicing events. Here, we present a simple approach to quantify the strength of individual splice sites, which determines their selection in a splicing reaction.Splice-site strength, as a quantitative phenotype, allows us to analyse splicing precisely in unprecedented ways. We demonstrate the power of this approach in defining the genomic determinants of the strength of individual splice-sites through GWAS. Our pilot-GWAS with more than thousand splice sites hints that cis-sequence divergence and competition between splice-sites and are among the primary determinants of variation in splicing among natural accessions of Arabidopsis thaliana. This approach allows deciphering the principles of splicing, which in turn has implications that range from agriculture to medicine.

The integral spliceosomal component CWC15 is required for development in Arabidopsis

“…The prevalent form of AS in animals is exon skipping 56 . We have utilized an approach that allows the detection of specific splice-sites that show aberrant usage between samples 40 . It has been estimated that 61% of Arabidopsis genes undergo AS, with the majority of AS events being intron retention 57 which in turn leads to nonsense-mediated decay (NMD) of the affected transcript 56 .…”

“…To analyze whether specific splice sites across the genome are affected in cwc15-1 mutants, we utilized SpliSER 40 , which enables the quantification of splicing at the level of individual splice sites. We first compared the variation in splice-site strength of the splice-sites across both tissues through PCA analysis.…”

“…A splice junction BED file was generated using RegTools v0.5.2 72 with the same intron limits. Each mapped RNA-seq sample was processed with SpliSER v0.1.1 and analyzed using the diffSpliSE pipeline 40 . To maintain the accuracy of the quantification, a splice site would be filtered out unless each replicate being assessed had at least 10 reads showing evidence of its utilization, or non-utilization.…”

Predicting RNA splicing from DNA sequence using Pangolin

“…We define the usage of a splice site as the proportion of transcripts of a gene that use a given splice site. SpliSER considers four types of reads to estimate usage for a target site: α and β 1 reads are split and non-split reads respectively that map to the target site; and β 2−SIMPLE and β 2−CRYPTIC reads are split reads that provide direct and indirect evidence against usage respectively (Dent et al, 2020). Then, SSE is calculated as: …”

“…Using sequences from multiple species has been shown to substantially improve prediction in some applications (Kelley, 2020), but has not yet been tested, to the best of our knowledge, in the context of RNA splicing (Supplementary Note 2). Specifically, we processed sequence and RNA splicing measurements from the four aforementioned species (Figure 1a, Supplementary Table 1), by using SpliSER (Dent et al, 2020) to quantify the usage of all splice sites after mapping RNA-seq data from heart, liver, brain, and testis from up to 8 samples per species per tissue (Methods). Next, we split genes into a training set and a test set.…”