Illumina launches SpliceAI2 for interpreting rare-disease splice variants
Published in Artificial Intelligence, Bioinformatics, Tools.
October 9, 2026 — Illumina announced SpliceAI2 on October 8, introducing an AI model for predicting how genetic variants affect RNA splicing. The release targets variant interpretation in rare-disease research and other genomic applications.
In analyses described in a newly released company preprint, the model identified 17% more disease-relevant splice variants than comparison models using matched DNA and phenotype data from Genomics England. On paired whole-genome and RNA sequencing data from GTEx, Illumina reported a 34% improvement in quantifying splice-site usage relative to the next-best model.
Those figures describe research benchmarks; they do not establish a corresponding increase in patient diagnosis rates or improved clinical outcomes. The preprint also remains distinct from peer-reviewed validation.
Illumina says training used a dataset 100 times larger than that of the original SpliceAI. Researchers can access the model through DRAGEN Annotation and Emedgene, connecting splice-effect predictions with existing genomic analysis workflows.