NEWS: Deepflare's Platform Validated in Prospective In Vivo Study

Lighting up Antigen Design with AI

We built the first AI platform to move beyond binding affinity and predict true T-cell immunogenicity. De-risk your pipeline, accelerate discovery, and achieve results with 12x greater accuracy.

OSIVAXCevaProgeneerUCSFBavarian NordicHawaii BiotechSagitta Biotech
DeepFlare protein design platform

Proven Results,
Trusted by the
Industry

Our platform was validated across multiple viral and cancer targets, and beat all of the state of the art models & solutions when solving real-world R&D challenges.

Hawaii BiotechProgeneerOSIVAX
12x

Higher Precision-Recall AUC in T-cell response prediction than the golden standard models.

See Case Study
3x

Higher precision demonstrated in a prospective, head to head, in vivo cancer vaccine study.

See Case Study
100%

Confirmed expression rate, as validated by prospective in vitro study on difficult viral target.

See Case Study
Try it now

Experience it: Go from target protein to de novo antigen design in seconds

DeepFlare protein design platform
Try it now in our interactive demo Try for free

Design next-generation antigen candidates.

Design, iterate, apply custom feedback and receive superior candidate within 24h.

Apply different criteria such as: increase or decrease immunogenicity, increase solubility, preserve or modify different regions.

Seamless Computational Workflow

Deepflare's platform integrates the most powerful tools in computational biology from open source (such as RFdiffusion) together with Deepflare's proprietary in-house models, into a single, seamless workflow.

Our Approach

Predict and mitigate the
immunogenicity risk

The Validation Crisis

The majority of antigen design models, such as netMHCpan 4.2, are benchmarked on biased data, creating a cycle of inflated metrics that seem to work well in silico, but fail miserably when tested in vitro & in vivo.

01The Validation Crisis

The Wrong Target

Antigen design models and products often mistake binding affinity for T-cell response. While binding affinity is needed for the latter to work, it is not synonymous. Especially when working in viral therapies, modeling these responses separately is crucial for the success.

02The Wrong Target

The Usability Barrier

There are countless bioinformatics & machine learning tools coming out every week, and it is impossible for innovative biotech's to implement or even test them all. We integrate the best of open source, with the addition on validated, proprietary Deepflare models.

03The Usability Barrier

Business model

You can test the platform online for free, using basic features. For premium features (modifying regions, optimizing immunogenicity) schedule a call with us to get started with SaaS platform access. For deeper integrations, we also work with enterprise customers, adding on top of the platform bespoke data training, added governance features, commercial exclusivity on targets & more.

04Business model

Get started for
free today

Gain 12x higher accuracy, intuitive bench-ready tools, and technology validated in prospective in-vivo studies.

Try for free