Explore how a DNA substitution changes a splice-site score.
A trained, compact research baseline compares a reference DNA window with a single nucleotide substitution. All inference runs on this host.
Research prototype. The model was trained to recognize splice junctions in an old mixed-primate dataset. Its score changes have not been validated against measured variant effects or clinical outcomes.
Loading model evidence…
Reference and alternate DNA
Enter exactly 60 bases in the transcript's 5′→3′ direction. The candidate junction is between bases 30 and 31. Positions below refer to this window, not genome coordinates. Reverse-complement minus-strand windows and alleles first.
Connect to this host
Research result
Run the public example or compare your sequence.
| Junction class | Reference | Alternate | Change |
|---|
These are uncalibrated classification scores. A positive donor or acceptor change does not establish a new splice site, disease risk, or pathogenicity.
Result provenance
No result yet.
Training and scope
240 position-specific nucleotide inputs → three splice-junction classes. Training, validation, and test windows are grouped by original sequence record and by at least 90% sequence identity to reduce leakage. This is not a chromosome or clinical cohort holdout.
Dataset: UCI Molecular Biology (Splice-junction Gene Sequences), 1991, CC BY 4.0. The data come from GenBank 64.1 and include primate sequences. Only single nucleotide substitutions in centered 60-base windows are supported.