Last updated August 24, 2026

Blank Bio

Training RNA foundation models to guide oncology trial patient selection.

OncologyPrecision medicineBioAI

RNA biology sits at the center of gene expression, disease mechanisms, and therapeutic response. Yet RNA programs are highly context-dependent, making it difficult to predict which interventions will work in specific patients or trials.

Blank Bio is training RNA foundation models for precision medicine and clinical trials. Its approach aims to use RNA data to better understand disease states, patient selection, and therapeutic response.

Key Investor Signals

Raised a $7.2M oversubscribed seed round in May 2026.

Blank Bio closed a $7.2 million oversubscribed seed round backed by Define Ventures, Leonis Capital, Nova Threshold, Ripple Ventures, SignalFire, and .

Partnered with PacBio to generate long-read RNA-seq data

Blank Bio and will generate HiFi long-read bulk RNA-seq data from up to 100 fresh-frozen patient tumor samples across multiple cancer indications at Seattle Children's Research Institute.

Selected for Y Combinator's Summer 2025 batch.

Blank Bio was selected for Y Combinator’s Summer 2025 batch, with Ankit Gupta as primary partner.

Other Positive Signals

  • Orthrus, the behind the company, was published in in April 2026. It outperformed existing genomic foundation models on mRNA property prediction while needing only a fraction of the fine-tuning data.
  • Team includes alumni from , , DeepMind, Amazon, MSK, Stanford, and .
  • Co-founder Philip Fradkin was co-first author on a NeurIPS 2024 paper on contrastive phenomolecular retrieval, which won best paper at the Foundation Models for Science workshop.

Team

Jonny Hsu

CEO & Co-Founder

  • Rose from Operations to Senior Strategy and Product Associate over three and a half years at Valence Discovery and , through Recursion Pharmaceuticals’ (Nasdaq: RXRX) acquisition in May 2023.
  • Built Polaris, an open benchmarking platform for machine learning in drug discovery, and Valence Labs’ public research portal.
  • Director of Investments at Front Row Ventures, Canada’s first university-focused venture fund, leading 17 associates across more than 30 Ontario campuses.
  • Analyst at digital-health venture firm Esplanade Ventures, after a science and business degree with a biochemistry specialization from the University of Waterloo.

Philip Fradkin

Co-Founder

  • One of the first employees at Deep Genomics, where he spent four years on machine learning for RNA biology.
  • Earned a PhD at the University of Toronto, with research affiliations at the Vector Institute.
  • Co-first author of Orthrus, the RNA foundation model published in Nature Methods in April 2026.
  • Co-first author of a NeurIPS 2024 paper on contrastive phenomolecular retrieval, best paper at the Foundation Models for Science workshop.

Ruian Shi

Co-Founder

  • Earned a PhD in computer science at the University of Toronto, with research affiliations at the Vector Institute.
  • Researcher in the Computational and Systems Biology Program at the Sloan Kettering Institute, .
  • First author of mRNABench, an open benchmark for mature mRNA property and function prediction.
  • Software engineer at Amazon, with earlier research-scientist internships at Amazon and Pinterest.

Product

Blank Bio is building RNA foundation models for oncology . Its platform analyses RNA-sequencing data to help pharmaceutical and diagnostic teams predict treatment response, assess disease progression, and improve existing molecular tests.

Built on existing RNA sequencing

Blank Bio uses bulk RNA-sequencing data already generated in oncology research and clinical development. Customers can apply the platform without changing how patient samples are collected or sequenced.

Functional RNA modelling

The underlying model learns which RNA sequences perform similar biological functions rather than predicting sequences one letter at a time. It is trained by comparing related genes across hundreds of mammalian species.

Three clinical applications

Predictive biomarkers help identify patients likely to respond to treatment. Prognostic scores estimate disease progression, while diagnostic partnerships apply the models to improve existing RNA-sequencing tests.

Market Overview

Oncology represents 41% of clinical trials, with 2,162 oncology studies started in 2024. These programs can remain in clinical development for more than a decade, making the selection of patients most likely to respond commercially important. Oncology programs using biomarkers for patient selection have historically reached approval at a much higher rate than programs without them.

RNA sequencing already captures molecular information from patient samples, but standard analysis often reduces this information to per-gene counts. This can discard transcript-level features that may help explain treatment response or disease progression. The opportunity sits in extracting more predictive value from an assay already used across oncology research and clinical development.

Oncology share of clinical trials
41%

Share of all clinical trials represented by oncology, with 2,162 studies started in 2024.

Approval likelihood with biomarkers
6.7×

Oncology programs using biomarkers for patient selection reached approval 10.7% of the time, compared with 1.6% without biomarkers.

Global cancer medicine spending
$252B → $441B

Projected growth from 2024 to 2029 at list prices.

Global RNA-sequencing market
$3.92B → $10.05B

Projected growth from 2024 to 2030, representing a 17.3% compound annual growth rate.

Comparable Outcomes

Oncology foundation model licensing

Noetik

$50M upfront plus annual licensing fees

built oncology foundation models trained on spatial-biology data to model gene expression, cell states and tumor-immune interactions. In January 2026, GSK licensed two Noetik models for lung and colorectal cancer under a five-year, non-exclusive agreement. The deal includes $50 million in upfront and near-term payments, plus annual subscription fees.

RNA expression signature

Genomic Health

IPO at ~$292M, later acquired for $2.8B

developed Oncotype DX, a breast-cancer test that analyses 21 genes to predict recurrence risk and likely benefit from chemotherapy. Genomic Health listed on Nasdaq in 2005 at $12 per share, valuing it at approximately $292 million. acquired the company in 2019 for $72 per share in cash and stock, valuing the transaction at $2.8 billion.

Transcriptome sequencing at scale

Caris Life Sciences

IPO at ~$5.8B valuation

Caris built a precision-oncology platform combining whole-exome and whole- sequencing with machine learning. It provides molecular profiling to clinicians and supports biopharma companies with treatment selection and drug development. Caris listed on Nasdaq in June 2025 at $21 per share. It sold 23.5 million shares for $494.1 million in gross proceeds, giving the company a valuation of approximately $5.8 billion.

Have a question about Blank Bio?

Ask DealBio. We may research the answer or share selected questions with the company.

Sources

This profile was built from public company materials, ecosystem sources, market references, and professional profiles.

Get the DealBio Pulse

Weekly insights on health and longevity startups, standout founders, key markets, and notable exits.

You can unsubscribe at any time.