Introducing CI-SpliceAI High Confidence Regions track
Last year, we introduced genome-wide precomputed CI-SpliceAI scores for VarSeq, enabling the detection of splice-altering variants outside of canonical splice motifs. These scores make it significantly easier to identify cryptic splice gains, splice losses, and other splicing events that may be missed by traditional algorithms.
We’re excited to announce a new companion track to CI-SpliceAI: the CI-SpliceAI High Confidence Regions track. This track solves a critical interpretation challenge that users face when working with splice predictions.
Interpreting Absence of a Score
The CI-SpliceAI track displays variants with a splicing impact score of 0.1 or greater, helping researchers identify genetic changes likely to disrupt normal splicing. However, when a variant is absent from the track, the interpretation has been ambiguous and, without additional information, this absence can mean one of two things:
- The variant has a low probability of impacting splicing
- No splice prediction scores have been computed for the variant, leaving its impact unknown
This ambiguity makes it difficult to determine whether a variant of interest has a low likelihood of impacting splicing.
CI-SpliceAI High Confidence Regions
The new High Confidence Regions track resolves this uncertainty by identifying genomic regions where we can confidently interpret the absence of a variant from CI-SpliceAI as indicating low splicing impact probability, provided certain criteria are met.
If your variant meets all of the following criteria, then it is unlikely to have an impact on splicing:
- Absent from the CI-SpliceAI track
- Overlaps a High Confidence Region
- Matches one of these variant types:
- Single nucleotide polymorphisms (SNPs)
- Single-base insertions
- Deletions of 1-4 base pairs
For other variant types, including large deletions of more than 4 base pairs, absence from CI-SpliceAI is not evidence of low splicing impact, even within High Confidence Regions.
Conclusion
This track empowers researchers to make more informed decisions when evaluating variants. Instead of requiring additional computational analysis for every variant absent from CI-SpliceAI, you can now quickly determine whether that absence indicates a low chance of splicing impact.
The CI-SpliceAI High Confidence Regions track is now available alongside the CI-SpliceAI track in the VarSeq. We encourage you to explore how these complementary tracks work together to provide a more complete picture of splicing impact across the genome. Please reach out to our team if you are interested in integrating the CI-SpliceAI annotation tracks into your VarSeq workflows.
