Skip to content

SpliceVault

Resource Location

/labs/SysMedBio/Q/Resources/Genomics/SpliceVault

/Volumes/lab/SysMedBio/Resources/Genomics/SpliceVault

Q:\SysMedBio\Resources\Genomics\SpliceVault

SpliceVault is a database and tool designed to predict the precise consequences of variants that disrupt canonical splice sites.

Scientific Context

Pre-mRNA splicing is the process by which introns are excised and exons are ligated to form mature mRNA. Genetic variants that disrupt canonical splice sites (typically the highly conserved GT and AG dinucleotides at exon–intron junctions) often lead to splicing defects. The consequences of these defects generally manifest as: 1. Exon skipping: Where one or more exons are omitted from the final transcript. 2. Cryptic splice site activation: Where splicing machinery utilizes alternative, previously dormant splice sites within an exon or intron.

While traditional predictive algorithms (such as SpliceAI) excel at determining whether a variant will disrupt splicing, they often fail to predict the precise structural outcome of the aberrant transcript (e.g., which specific exon will be skipped or which cryptic junction will be used).

SpliceVault solves this by leveraging empirical RNA-sequencing (RNA-seq) data. The tool works on the biological principle that "natural" splicing errors (low-frequency alternative splicing events that occur baseline in healthy tissues) represent the exact alternative pathways that will be activated if the canonical splice site is mutated. By compiling and ranking the "Top-4" alternative splicing events across a massive dataset of over 335,600 human RNA-seq samples, SpliceVault provides highly accurate, pre-calculated predictions of the exact mis-splicing outcome. It correctly identifies ~96% of exon-skipping events and ~86% of activated cryptic splice sites in clinical cases, providing invaluable utility for classifying variants of uncertain significance (VUS).

Main Publication

  • Title: SpliceVault predicts the precise nature of variant-associated mis-splicing
  • Authors: Ruebena Dawes, Adam M. Bournazos, Samantha J. Bryen, Shobhana Bommireddipalli, Rhett G. Marchant, Himanshu Joshi, and Sandra T. Cooper
  • Journal: Nature Genetics (2023, Volume 55, Pages 324–332)
  • DOI: 10.1038/s41588-022-01293-8

Available Files

The SpliceVault directory contains precomputed variant mis-splicing outcome tables:

  • SpliceVault_data_GRCh38.tsv.gz: Precomputed splice-site variant predictions mapped to the GRCh38/hg38 human assembly.
  • SpliceVault_data_GRCh38.tsv.gz.tbi: Tabix index file enabling fast coordinate queries for the GRCh38 dataset.
  • SpliceVault_data_hg19.tsv.gz: Precomputed splice-site variant predictions mapped to the GRCh37/hg19 human assembly.
  • SpliceVault_data_hg19.tsv.gz.tbi: Tabix index file enabling fast coordinate queries for the hg19 dataset.