Aggregation
Does it clump or fragment?
SEC · CE-SDS
Monomer purity and fragmentation from size-exclusion and capillary electrophoresis.
Antibody Developability
Find liabilities before they cost you a program.
Thousands of designs. More than a dozen assays. Data built for machine learning.
- designs per batch
- 2,400+
- designs per batch
- biophysical assays
- 12+
- biophysical assays
- best-in-class production, or CHO/HEK
- CFPS
- best-in-class production, or CHO/HEK
- for you or us to model on your behalf
- AI-ready
- for you or us to model on your behalf
Any antibody format
IgG, VHH, VHH-Fc, Fab, scFv, minibinders, multi-specifics, and ADCs, with run-to-run consistency validated against published literature.
What the panel tells you
5 risk areas
Thermostability
When does it unfold?
nanoDSF
Melting and onset temperatures to rank conformational stability.
Hydrophobicity
Is the surface sticky?
HIC · SMAC
Hydrophobic interaction and colloidal stability chromatography.
Self-association
Will it stick to itself at high dose?
AC-SINS · DLS-kD
Self-interaction signals that predict viscosity and high-concentration behavior.
Binding & specificity
Does it bind only what it should?
BLI · CHO/Ova · HAC
Target and FcRn binding, polyreactivity, and heparin binding.
Also available: titer, formulation screening (FaST), and functional assays.
[ Offerings ]
Three entry points into one discovery funnel
Start at any stage
- [ Find binders ]Stage 1
Cell-free binder screening
Rapid expression and binding screens for minibinders, VHHs, and scFvs.
You receive
Screened hits for follow-on work
See Stage 1 ↓ - [ Select candidates ]Stage 2
Candidate profiling
Binding, cell-based function, and developability across a campaign to rank leads.
You receive
Integrated candidate profiles
See Stage 2 ↓ - [ Build models ]Stage 3
Large-scale data & modeling
Standardized developability data across thousands of antibodies.
You receive
AI-ready datasets and model results
See Stage 3 ↓
[ Stages 1–2 · Lab-in-the-loop ]
Built for AI-driven design campaigns
You bring AI-designed panels. We build and test each cycle fast. Then advance the top binders to full function and developability profiling.
Stage 1 — iterative discovery
E. coli CFPS · fast TATDesign
Your AI models
Build
Cell-free expression
Test
BLI binding + reporter function
learn + redesign over multiple cycles
Fast loop first: every cycle scores binding and reporter function before any CHO developability work.
Top binders
Stage 2 — function + developability
CHO · VHH-FcReformat to VHH-Fc
Half-life & clinical assay format
Express in CHO
Mammalian production of the lead panel
Developability + function
PROPHET-Ab panel & richer functional assays
[ Stage 3 · AI/ML ]
Stage 3 — large-scale data & modeling
Data and compute built for developability models.
We combine scalable Google Cloud compute and shared model-training infrastructure with protein-specific ML expertise and standardized, high-throughput assay data.
Our previous work showed improved prediction as training data expanded, while the consortium provides a path to harmonized data and federated model development across partners.
[ Consortium ]
Antibody Developability Consortium
AbbVie, argenx, Lundbeck, and Takeda are building a 10,000-antibody federated dataset with Ginkgo and Apheris. We are building the global model together.
Antibody Developability — Service Overview
Assay specs, validation data & timelines