Massive Parallel Splicing Assay identified common exonic and deep intronic variants to be splicing associated variants in myopia and age-related macular degeneration
Introduction
Genome-Wide Association Studies (GWAS) and Whole Exome Sequencing (WES) have unveiled numerous genetic variants, with many remaining functionally uncharacterized. Historically, research has focused on missense mutations impacting protein function and variants influencing enhancer activity. However, recent studies reveal that variants affecting RNA splicing may play an equally crucial role, especially in disease mechanisms (Nature Communications, 2022).
The technology for predicting the effects of variants on splicing has advanced significantly in recent years through in silico methods. Among the tools available, SpliceAI stands out as a state-of-the-art model, though other algorithms, including MMSplice, SQUIRLS, and ConSpliceML, also show competitive and, in some cases, superior performance. However, high predictive accuracy often does not translate to clinical diagnostics. For instance, while SpliceAI’s precision-recall area under the curve (AUC) reaches 0.98 in theoretical datasets (Jaganathan et al., 2019), its performance diminishes in real-time clinical testing (Ellingford et al., 2019; Wai et al., 2020).
Minigene assays, the traditional gold standard, provide direct observations but are resource-intensive and impractical for analyzing thousands of variants (Human Mutation, 2022). Massive Parallel Splicing Assay (MPSay) offers a scalable alternative, allowing multiple variants to be synthesized and tested simultaneously, effectively functioning like thousands of minigene assays at once. Initially applied to rare exonic variants, we further used MPSay to assess whether common variants could exhibit splicing-altering effects comparable to those observed in rare variants, thus extending its application to common exonic variants.
Given that most GWAS variants reside in deep intronic regions and evidence shows these variants can disrupt splicing by interacting with intronic splicing enhancers or silencers, we aimed to extend detection capabilities to these deep intronic regions. Recognizing that deep intronic mutations often alter splicing by introducing new acceptor sites, we developed a targeted approach, the Massive Parallel Splicing Assay for Splicing Acceptor (MPSac), designed specifically to identify splicing-associated variants. This enables testing across both common and rare GWAS and WES variants, potentially establishing MPSac as a valuable tool for elucidating the roles of intronic variants in splicing.
Two ensemble learning methods, adaptive boosting and random forests, were used to construct models that take advantage of individual methods.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4267638/
MaxEntScan, a tool that predicts the 5′/3′ splice site
https://pubmed.ncbi.nlm.nih.gov/15285897/
0.708 adaboost
randomforest 0.515