Model Check
From $5KTest the model you already have against a private, experimentally measured holdout set.
- Frozen assay and controls
- Predictions locked before testing
- Rank correlation + failure analysis
- Raw data and model-ready CSV
Sequence in · ML-ready experimental data out
Turn protein sequences into biological evidence.
Upload 24–1,000 designs. Get model-ready expression, binding, stability and developability data—including full-length glycosylated IgG. Initial expression results target 3–4 weeks.*
START WITH AN EXAMPLE
No account required to start. A Genesis scientist reviews every campaign before scope, timing or pricing becomes a laboratory commitment.
3–4 weeks*initial expression results
$499full IgG developability
Negative dataincluded by default
*Indicative turnaround begins after technical acceptance and receipt of complete inputs; sequence count and length, construct format, materials, method readiness, execution site and capacity can change timing. Published pricing is also subject to review.
MODEL DATASET / GBF-024● IN REVIEW
Target areaPD-L1 extracellular domain
Designs48
Panel7 assays
FormatFull IgG
OutputCSV + RAW
01AcceptedSequence & antigen review
02GenerateExpression & purification
03MeasureFixed assay panel
04DeliverAnalysis-ready dataset
01 / Scale & pricing
Start by proving the data changes a decision.
The lowest-friction entry for a foundation-model team is a held-out Model Check: freeze predictions, measure a private cohort, and learn exactly where current performance holds or fails. Scale only after the first dataset earns it.
24–48 held-out designs
Decision Dataset
From $15KGenerate enough comparable data to select candidates, diagnose model errors, or establish a benchmark.
- Multi-property assay panel
- Positive and negative outcomes
- Scientist-designed controls
- Reusable private dataset
Learning Loop
CustomReserve capacity for repeated design–build–test–learn rounds or a proprietary data program.
- Repeated active-learning rounds
- API and campaign workspace
- Reserved laboratory capacity
- Custom methods and rights
THE $5K MODEL CHECK
01
→Submit a model, scoring function or ranked candidate set.
02
→Freeze predictions on a scientist-reviewed holdout cohort.
03
→Receive experimental results, rank metrics and failure analysis.
04
Decide whether a larger dataset can improve the model.
Screen
$149/proteinExpression + binding
Expression results in 3–4 weeks*
Core
$229/proteinExpression + binding + thermostability
Expression results in 3–4 weeks*
Developability
$389/proteinSeven co-measured assays
Expression results in 3–4 weeks*
Developability — IgG
$499/proteinSeven assays · mammalian · glycosylated
Expression results in 3–4 weeks*
250+ −10%500+ −20%1,000+ −25%Open benchmark −20%
*The 3–4 week expression target begins after technical acceptance and receipt of complete inputs. Sequence count and length, construct format, materials, method readiness, execution site and capacity can change timing. Volume and open-data discounts stack. At 1,000 published IgG records, the current launch schedule is approximately $300 per antibody. Prices exclude unusual target sourcing, custom methods and nonstandard material requirements.
THE FULL-LENGTH ADVANTAGE
Do not stop at “does it bind?”
Cell-free screens are excellent for fast small-format validation. Genesis extends the same sequence-to-data experience into the development questions that require mammalian expression, glycosylation and full-length antibody context.
CAPABILITYSMALL-FORMAT SCREENGENESIS FULL IgG
Binding kineticsIncludedIncluded
ThermostabilityIncludedIncluded
SEC + HICAdd Genesis panelIncluded
Polyreactivity + ELISAAdd Genesis panelIncluded
Mammalian glycosylation—Included
Negative outcomesRetainedRetained
02 / Ways to work
Choose one way to get started.
These are three entry points, not sequential steps. Each ends in a defined experimental record: raw files, analyzed values, metadata, QC and labeled failures.
Upload sequences
Choose a published bundle for a defined model-validation or developability cohort.
Plan a data campaign
Describe the target, model objective and decision. We draft the cohort, panel, controls and schema.
Partner with the foundry
Custom assays, repeated active-learning rounds, private datasets and product-development programs.
03 / Campaign workbench
Move from an idea to a technically reviewable request.
This first version prepares the scope. A Genesis scientist confirms feasibility, sample requirements, capacity, timing and final pricing before any SOW is issued.
DATA RIGHTS
Customer project data remains customer-owned by default. Shared benchmarks or reusable dataset rights require a separate opt-in agreement.
CAMPAIGN BRIEF
SCIENTIST REVIEWEDWhat decision should the data support?
04 / Measurement layerOne cohort. One protocol.
One cohort. One protocol.
Every useful outcome retained.
Choose individual assays or a fixed bundle. Method readiness, material requirements and capacity are confirmed during technical acceptance.
COMING SOON
Expression + yield
Standalone expression service. Expression remains scoped within accepted multi-assay campaigns.
$99
Binding kinetics
BLI: KD, kon, koff and sensorgrams
$99
Thermostability
Tm, unfolding curves and replicate QC
$89
SEC aggregation
% monomer, %HMW and chromatograms
$89
HIC hydrophobicity
Relative retention and run-level controls
$109
Polyreactivity
Specificity risk under a frozen method
$79
Functional ELISA
Target-specific functional readout
+$149
Full-length IgG
Mammalian expression and glycosylated context
DELIVERY SCHEMA
Built for analysis, not trapped in a report.
Every accepted campaign includes the experimental context needed to interpret the result and reuse it later.
- Raw instrument files
- Analyzed results
- Replicate-level values
- QC and failure codes
- Construct and run metadata
- Human-readable data dictionary
campaign_results.csv3 / 48 rows shown
| ID | FORMAT | KD nM | % MON | Tm | QC |
|---|---|---|---|---|---|
| GBF-001 | IgG1 | 2.4 | 98.1 | 72.3 | PASS |
| GBF-002 | IgG1 | 8.7 | 95.4 | 68.9 | REVIEW |
| GBF-003 | VHH | 0.9 | 99.2 | 75.1 | PASS |
RAW ATTACHMENTS
sensorgrams/ · chromatograms/ · plate_exports/ · protocol.json · data_dictionary.md
05 / Two buyers, one data engineVolume for models.
Volume for models.
Decisions for pharma.
Foundation-model campaigns build comparable training data and benchmarks. Pharma programs use the same workflow to triage full-length antibody candidates.
Foundation-model teams
- Developability-prediction training sets
- Binder validation with negative data
- Cross-model comparison on one protocol
- Public benchmark contributions
- Antibody-format extension
Pharma biologics teams
- Lead-candidate triage
- Early developability gating
- Biosimilar and comparability screens
- AI-designed candidate validation
- IgG, Fab and scFv format selection
SCIENTIST-REVIEWED SCOPING
Turn one model question into a testable campaign.
Book 25 minutes to review the biological objective, candidate cohort, current model performance, assay panel and the smallest useful pilot. The call produces a next-step recommendation—not an automatic laboratory commitment.
25 minutesGoogle Calendar connectedCal Video
Open booking page ↗06 / Provenance
A new interface, backed by an established diagnostics operation.
Santa ClaraSatellite laboratory for eligible US-executed programs
AsiaGenesis IVD laboratory and manufacturing operation
20+ yearsDiagnostics R&D and assay development
550,000 ft²Factory area reported by Genesis IVD ↗
ISO systems13485 and 9001 quality discipline at Genesis IVD
Plate-enabledELISA, BLI, HPLC and FACS capabilities scoped by site
Genesis BioFoundry is a laboratory-data service being developed with Genesis IVD as the initial laboratory based in Asia and a satellite laboratory in Santa Clara, California. The executing site, method readiness, materials, data-location requirements, capacity and applicable quality requirements are confirmed during technical review. Genesis BioFoundry is not presented as a separately accredited clinical facility.
What data does your model need next?
Bring sequences, a target area or a candidate-ranking problem. We will turn it into a technically reviewable campaign and a reusable experimental record.