NovaLab

Where AI meets biology.

NovaLab is Geodesic's in-house wet-lab, built to connect our AI models directly with the physical world through rapid experimental validation and iteration.

Experimental workflow

From designed binder to validated therapeutic.

Every AI-designed binder entering NovaLab moves through a structured experimental workflow designed to answer progressively harder questions: Can we make it? Is it what we designed? Does it bind? How well does it bind? Does the signal hold up across assays? Can it become a viable therapeutic? And ultimately — does it work?

01

Expression & Purification

Can we make it?

AI-designed sequences are expressed recombinantly and purified in-house, turning computational designs into physical molecules ready for testing. Antibody fragments and designed binders are captured by affinity tag; full-length IgG is purified via Protein A.

02

QC Gate

Did we make the right molecule?

Purified proteins clear a QC gate on yield and concentration, reduced and non-reduced SDS-PAGE, and SEC-HPLC monomer purity before advancing into downstream characterization.

03

Affinity

Does it bind?

Designs are screened against their intended targets by SPR, at single or dual concentrations, to triage binders from non-binders across the full library.

04

Binding Kinetics

How does it bind?

Promising hits move into multi-concentration SPR kinetics, a dilution series with replicates that resolves the full association and dissociation behavior: ka, kd, and KD.

05

Orthogonal Validation

Is the signal real?

Key results are re-tested by BLI in reverse binding orientation, confirming the signal holds on an independent platform and isn't an artifact of one assay geometry.

06

Developability Panel

Can it become a drug?

Validated binders are profiled for thermal stability (Tm), polyreactivity, and off-target cross-reactivity, the biophysical properties that most often separate a good binder from a viable therapeutic.

07

Functional Assay

Does it work?

The strongest candidates advance beyond binding into functional assays designed to measure the biological activity that matters for the intended therapeutic mechanism.

The loop

Design, experiment, learn, repeat.

Computational designs move rapidly into experiments. Experimental results generate high-quality biological data. What we learn flows back into the next generation of models.

next generationof modelsAI modelsDesignExperimentLearn
Closing the loop

Computational intelligence, meet biological reality.

By tightly integrating AI and experimentation, NovaLab closes the loop between computational intelligence and biological reality. Our goal is to make every cycle faster, more informative, and more capable — accelerating the path from biological ideas to new medicines.