Life sciences · Bio pipelines
Run bio tools as pipelines, not setup projects.
MMseqs2, Foldseek, US-align, ProteinMPNN and ESM-2 are ready to run from a form or a REST call. Connect them into a pipeline, fan out over many inputs, and keep a record of every run.
One pipeline, four tools.
A design and triage pipeline hands each tool’s output to the next, on the machines each one needs.
Design and triage pipelineProteinMPNN
- design
- GPU × 1
MMseqs2
- search
- 16 vCPU
Foldseek
- structure
ESM-2
- train
- GPU × 1
What you get.
Public bio tools, ready to run
Search sequences, compare structures, design proteins and train predictors without installing each tool.
Pipelines and batches
Connect tools into a pipeline, fan out over many inputs, and rerun only the part that needs it.
Data stays where it must
On your infrastructure, bio jobs have no outside network access by default and reach only your object storage and DNS.
Results you can reproduce.
Each bio job keeps a record of how it ran, and shows its cost before you submit it.
A record of every run
The tool definition, image digest, settings, input checksums and resources used.
Costs before you run
See an estimate before you submit, then the actual usage in the same unit when the job ends.
Agents with approval
Connect an agent over MCP. It can find tools and request runs, and work starts only after a person approves.
Job record
- Tool definition
- proteinmpnn · 3f9a1c07
- Image digest
- sha256:8b2e6d…41d0
- Settings
- num_sequences 16, sampling_temp 0.2
- Input checksum
- sha256:c07a15…9e12
- Estimated usage
- 1.2 CU
- Actual usage
- 1.1 CU
- Approved by
- Project admin
Questions.
Can we add our own tools?
Yes. Register a Python script or a container image from an allowed registry and run it like the public tools.
Can we call it from code?
Yes. Every tool and pipeline runs from a REST call as well as from the form.
Where does our data go?
In our cloud it stays in Supergate Cloud; on your infrastructure it stays there. Access is set out in your agreement.