Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Illumina released SpliceAI2 on October 8, 2026. It is a genomic AI model that predicts how DNA sequence shapes splicing: which splice sites a cell uses and how often, how those sites connect into junctions, and which full-length transcript isoforms result. Illumina positions it as a tool for studying the functional effects of genetic variation, particularly splice variants relevant to rare-disease research. The launch materials label it “For Research Use Only” and “Not for use in diagnostic procedures.”
What SpliceAI2 predicts
Illumina describes SpliceAI2 as covering three linked outputs. Each one answers a different question about a variant’s effect on the transcript.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
Genetics: A Conceptual Approach | $71.34 | Buy on Amazon |
| 2 |
|
Genetics: A Conceptual Approach | $236.15 | Buy on Amazon |
| 3 |
|
Genetics For Dummies | $16.14 | Buy on Amazon |
| 4 |
|
Genetics: From Genes to Genomes, 5th edition | $141.57 | Buy on Amazon |
| 5 |
|
Thompson & Thompson Genetics and Genomics in Medicine (Thompson and Thompson Genetics in Medicine) | $68.29 | Buy on Amazon |
- Splice sites: which positions a cell uses for splicing, and how frequently each one is used.
- Junctions: which splice sites connect to one another to form a junction.
- Transcript isoforms: the full-length transcripts that result from those connections.
The practical motivation Illumina gives is that sequence-based predictions could help researchers judge the likely consequences of a variant on transcripts without first obtaining RNA from the tissue where the gene is expressed. Illumina presents this as a model goal. It does not claim that sequence-only prediction replaces experimental validation in every case.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →How SpliceAI2 differs from the original SpliceAI
Illumina describes the original SpliceAI as focused on whether a cell splices at a given position. SpliceAI2 extends that scope from individual sites to junctions, complete transcripts, and tissue context. The table below sets out the differences as Illumina describes them in its October 8, 2026 research article, “Introducing SpliceAI2: The Next Generation of Splicing and Transcript Isoform Prediction.”
#1 Best Overall
| Question the model addresses | Original SpliceAI (as described by Illumina) | SpliceAI2 (as described by Illumina) |
|---|---|---|
| Whether a site is used at a given position | Core focus | Yes, with usage frequency |
| Which splice sites connect into junctions | Not stated | Yes |
| Which full-length transcript isoforms result | Not stated | Yes, trained with 330 ENCODE long-read samples |
| Tissue-specific splicing patterns | Not stated | Yes, across 48 human tissues, with a stated limitation (see below) |
| Training data scale | Not stated in the launch material | 314,745 RNA-sequencing samples spanning ten species |
How the model was trained
Short-read RNA-sequencing data
Illumina reports that the SpliceAI2 training set included 314,745 RNA-sequencing samples across ten species. After filtering, it contained more than 46 million observed splice junctions. These figures come from Illumina’s launch article and describe the training set, not the model’s accuracy.
Long-read data for full transcripts
Short reads capture individual junctions well, but they make it harder to link splicing events across a whole transcript. For that reason Illumina added 330 ENCODE long-read samples to training, so the model can connect splicing events across an entire transcript. For genes that were not seen during training, Illumina reports that SpliceAI2 reconstructed the most common transcript 82% of the time with long-read training, compared with 78% using short-read data alone. Those figures are Illumina’s and apply to the gene set it tested.
Rank #2
Reported performance and what the numbers cover
Every performance figure in the launch comes from Illumina’s own analyses. The table lists each figure with the conditions Illumina attaches to it.
| Comparison | Reported result | Conditions |
|---|---|---|
| Disease-associated variants identified, SpliceAI2 vs. other tested splice models | 17% more | At matched confidence thresholds; Illumina analysis of phenotype and DNA data from 7,504 Genomics England participants. The October 8, 2026 press release describes this as a rare-disease research dataset. |
| Disease-relevant splice variants, SpliceAI2 vs. legacy SpliceAI | 33% more | At a 2X confidence interval; same analysis |
| Disease-relevant splice variants, SpliceAI2 vs. legacy SpliceAI | 66% more | At a 4X confidence interval; same analysis |
| Cryptic splice variants identified by SpliceAI2 | Roughly 50% deep intronic | Same analysis; the share of cryptic variants located deep in introns |
| Transcript reconstruction for genes not seen in training | 82% with long-read training vs. 78% with short-read only | Illumina launch article; most common transcript |
Four points determine how these numbers should be read:
Rank #3
- They are vendor-reported. The launch materials contain Illumina’s benchmarks, and no independent reproduction accompanies them.
- The comparators differ. The 17% figure compares against other tested splice models, while the 33% and 66% figures compare against legacy SpliceAI at different confidence intervals. The three figures are not interchangeable.
- The benchmark set is specific. Illumina compared SpliceAI2 with original SpliceAI, Pangolin, and AlphaGenome across three benchmark datasets. The AlphaGenome comparisons were run by University of Oxford academic collaborators. Illumina states that SpliceAI2 performed best across its tested benchmarks, including for variants that create new splice sites, especially deep intronic variants.
- They are not diagnostic yield. The percentages describe what SpliceAI2 flagged in one analysis. They do not measure how often the model would lead to a clinical diagnosis or how accurate it is in general.
Tissue-specific predictions and their limits
Illumina reports tissue-specific results across nearly 15 million splice-site differential-usage measurements in 48 human tissues. Differential usage describes how a splice site’s use changes between conditions, here between tissues. The same article cautions that the model was less successful at predicting how the effect of a particular variant changes from one tissue to another. The tissue’s baseline splicing program was a stronger signal than the variant-specific shift. In practice, SpliceAI2 is better described as capturing tissue-level splicing patterns than as reliably predicting tissue-dependent variant effects.
Access routes and intended use
- BioInsight Platform applications: Illumina names DRAGEN Annotation and Emedgene as access routes.
- Illumina Connected Insights: named in the detailed launch article as a further route.
- Public repository,
Illumina/SpliceAI2on GitHub: includes source code, trained models, and precomputed predictions for possible single-nucleotide variants within human gene bodies, along with population-observed indels.
Access details can change, so confirm the current product documentation and the repository’s README before building a workflow around any of these routes. The launch page labels the material “For Research Use Only” and states “Not for use in diagnostic procedures.” The launch does not authorize clinical diagnostic use.
Rank #4
How to compare SpliceAI2 with other splicing tools
The launch supplies vendor-reported comparisons but no complete independent head-to-head review. When you read a comparison of SpliceAI2 with any other tool, check these axes first:
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Prediction target: a single splice site, a junction, or a full transcript isoform.
- Benchmark dataset and cohort: the named cohort, its size, and its disease context.
- Threshold and comparator: the confidence threshold used and the exact version of the competing model.
- Deep intronic changes: whether the result covers variants far from exons, which Illumina reports as a strength.
- Tissue context: whether the claim is about shared splicing patterns or about how a variant’s effect differs by tissue.
- Evidence type: whether the result is an in-silico prediction or an experimentally observed splicing outcome.
What Illumina said at launch
In the October 8, 2026 press release, “Illumina releases SpliceAI2 to help advance rare disease research,” two Illumina executives described the model’s role:
Best Value
- Rami Mehio, senior vice president and general manager of BioInsight at Illumina, said: “Variant effect prediction tools, such as SpliceAI2, are among the key areas of focus for the BioInsight AI Lab.”
- Kyle Farh, vice president of Illumina’s BioInsight AI Lab, said: “Illumina is advancing AI to systematically shrink the portion of the genome that remains uninterpretable.”
”
The Bottom Line
SpliceAI2 extends splice-site prediction to junctions and full transcript isoforms, and Illumina’s strongest supporting evidence is its training scale and its long-read transcript results. Its headline performance figures come from Illumina’s own cohort analysis, so they are best treated as a promising, vendor-reported signal to test in your own datasets. Because the launch is labeled for research use only, it belongs in research workflows rather than in clinical reporting.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
