Solutions
For developers
Everything the interface does, the API does. Scoped keys, asynchronous jobs, webhook callbacks and standard formats — so the platform becomes a step in your pipeline instead of a destination.
API
for every engine
Webhooks
job callbacks
Scoped
API keys
Reserved
GPU capacity
01
The problem
Scientific software has a habit of being a destination. You log in, you do the work by hand, you download a file, and you carry it somewhere else. That is tolerable for an occasional analysis and intolerable as a step in a pipeline that is supposed to run without anyone watching it.
The usual workaround is to reimplement the parts you need, which means maintaining a GPU environment, tracking model versions, and rebuilding the reproducibility machinery that made the results worth having. Most teams would rather call something.
02
What the API gives you
Scoped API keys
Keys are issued per workspace with explicit scopes, so a key that submits jobs does not also manage billing. Keys are rotatable and individually revocable, and usage is attributed per key.
Asynchronous by default
Submit a job, get an identifier, poll or wait for the callback. A full ESM-2 matrix and a one-second codon adaptation go through the same interface, which keeps client code uniform.
Webhook delivery
Register endpoints for job completion, batch completion, export generation and validation results. Deliveries are signed and retried, so a brief outage does not drop a result.
All of them, not a subset
Flux analysis, variant scanning, folding, docking, guide design and off-target sweeps, codon optimization, primers, annotation, assembly, phylogenetics and export are all reachable.
Formats that travel
GenBank, FASTA, SBOL and ST.26 for sequences; JSON for everything structural; CSV for tables. Catalog metadata — 1,000 constructs, 65 organisms, 70 series — is queryable directly.
Reserved GPU
Enterprise workspaces can run against reserved capacity rather than the shared queue, which is what makes a nightly batch finish on a predictable schedule.
03
A representative integration
- Trigger
- New sequence lands in internal LIMS
- Call
- Annotate, then submit ESM-2 scan
- Mode
- Asynchronous, webhook callback
- Follow-up
- Fold shortlist, Standard GPU class
- Assembly
- Construct built via API, CPU, free
- Validation
- Run before export, result delivered by webhook
- Delivery
- Export package pushed to internal store
A pipeline nobody has to sit and watch. A sequence arrives, gets annotated and scanned, the shortlist gets folded, a construct is assembled and validated, and a package lands in internal storage with its manifest. The team sees it when the validation report flags something worth a human decision.
Note where the money goes. Annotation, assembly, validation and export are CPU and cost nothing however often the pipeline fires. Only the scan and the folds consume credits, which makes the running cost of an integration like this easy to predict in advance.
04
Getting started
Any account can issue API keys and drive the CPU engines for free, which is enough to build and test a full integration before spending anything. GPU calls draw on the same prepaid credits as the interface — $25 for 250 credits, one credit for roughly one Standard job, with Performance and Max classes selectable per request.
Enterprise workspaces add reserved GPU capacity, higher rate limits, SSO, and webhook configuration managed at the workspace level rather than per key. That is the tier most teams move to once the integration stops being an experiment.
Ready to run it
Try this in the Splice Platform.
CPU engines are free to run. GPU engines run on prepaid credits.
Related