Pathogenic tandem splice acceptor variants escape screening by in silico tools

Authors

• PV Daude
• R Ramasawmy
• A Javed
• H Xue
• P Kellman
• K Chow
• AE Campbell-Washburn

Abstract

Introduction. Tandem splice acceptor (NAGNAG) sites are a known mechanism of alternative splicing and serve as loci of transcriptional regulation. Previous literature demonstrated that variants modifying existing, or creating novel, NAGNAG sites can cause mendelian disease. Despite advances in in silico splicing prediction tools, predicting the pathogenicity of NAGNAG variants remains difficult and, as a result, are rarely contextualized in clinical reporting.

Methods. We characterized NAGNAG variants in reference databases and among 1236 individuals from the NIH Undiagnosed Diseases Program (UDP). We evaluated pathogenicity of each NAGNAG variant in 130 UDP cases with RNA-Seq data.

Results. We found a depletion of AG-containing trimers from position -3 to -14, which often overlaps acceptor site enhancer motif. SpliceAI scores of AG-gain variants in that region were similarly high in UDP and reference populations. Evaluation of RNA-Seq data in UDP cases showed that variants with verified usage were more abundant in the -3 to -14 window compared to the rest of the intron. Furthermore, while those variants verified tended to have high SpliceAI acceptor gain scores, they had a PPV of 0.344, demonstrating the incomplete ability of in silico tools to assess AG-gain variants.
Conclusions. The pathogenicity of variants affecting splicing are difficult to assess. We show that a subset of those variants, those which create NAGNAG sites, frequently lead to the usage of a novel acceptor site. Predicting the function of those variants using in silico tools is inconsistent. This supports the use of empirical evaluation of AG-gain variants by RNA-Seq.

Scientific Focus Area: Genetics and Genomics

This page was last updated on Monday, September 25, 2023