PacBio and TychoBio Will Train RNA Models on 10000 HiFi Samples
PacBio and TychoBio announced a collaboration on 6 October 2026 to generate full length RNA data from more than 10000 samples to train models for steric blocking antisense and future siRNA design.
What changed
On 6 October 2026 PacBio and AI drug discovery company TychoBio announced a collaboration to generate full length RNA data for model training in RNA therapy design. TychoBio plans to use PacBio HiFi sequencing and Kinnex long read RNA sequencing on more than 10,000 samples, building datasets that link design of steric blocking antisense oligonucleotides (SBOs) to transcriptome wide effects, with plans to extend the approach to siRNA and other RNA targeting compounds.
TychoBio will test candidate RNA therapies and use PacBio sequencing to measure how each compound changes the transcriptome across multiple cell types, feeding results into models meant to predict efficacy, duration, and off target risk.
TychoBio founder and CEO Felix Raimundo said in the release that AI models are only as good as training data, and that HiFi long read sequencing enables full length transcript views including isoform and splicing changes relevant to rare diseases.
Why it matters
RNA therapeutic design still bottlenecks on sparse, short read training sets. A 10,000 sample HiFi program is a data moat play: whoever owns the transcriptome response library may shorten iteration cycles for SBOs and siRNAs in rare disease.
For pharma BD teams, the deal type matters as much as the science: sequencing vendor plus AI biotech, with explicit expansion from SBOs to siRNA.
Who is affected
Rare disease RNA therapy developers, computational biology leads choosing training corpora, and PacBio customers weighing Kinnex for functional genomics beyond germline sequencing.
What to do next
If you partner on RNA programs, ask whether your discovery stack can ingest full length isoform shift labels per compound, not just target knockdown summaries.
What to watch
First public model metrics from TychoBio tied to the 10,000 sample tranche, and any expanded modality claims beyond SBOs into siRNA as described in the October announcement.
Sources
- Primary. PacBio via GlobeNewswire, PacBio and TychoBio Collaborate to Train AI Models for Rare Disease RNA Therapeutics (6 October 2026). States sample scale, modalities, and CEO quote.