canonical: https://jentic.com/apis/googleapis.com/lifesciences

# Google Cloud Life Sciences API

The Cloud Life Sciences API runs container-based pipelines on Google Cloud, primarily for genomics, biology, and other life sciences batch workloads. It exposes a pipelines:run operation that schedules a Pipeline definition onto a Compute Engine VM and a set of operations endpoints to list, get, and cancel the resulting long-running operations. Locations can be discovered via the locations:list endpoint to choose a regional endpoint for compliance.

## For AI agents

Run container-based genomics and bioinformatics pipelines on Google Cloud and track or cancel the resulting long-running operations.

## Scope

Does not store genomic data, train ML models, or expose Compute Engine VM management - use only to run, list, get, and cancel container-based pipeline operations.

## Capabilities

- Run a containerized pipeline by submitting a Pipeline definition to pipelines:run
- List long-running operations to track pipeline progress in a region
- Get the status, metadata, and result of a single pipeline operation
- Cancel a running pipeline operation
- List supported regional locations to choose a compliant endpoint

## Use cases

### Run a genomics variant-calling pipeline

Bioinformatics teams package GATK or DeepVariant tools into a container and submit them via POST /v2beta/{+parent}/pipelines:run. Cloud Life Sciences allocates a Compute Engine VM with the requested machine type and disks, runs the pipeline, and returns a long-running operation that the team can poll for status.

Example prompt: Submit a Pipeline with one Action running gcr.io/myproj/deepvariant:latest and read the returned operation name.

### Operation polling and status reporting

Long-running pipelines complete asynchronously, so a workflow orchestrator must poll GET /v2beta/{+name} until done becomes true. The metadata block reports event timelines and any errors so an agent can report progress to the data scientist or trigger downstream steps.

Example prompt: Poll GET /v2beta/{+name} every 30 seconds until done=true, then return metadata.events for human review.

### Cancel a runaway pipeline

When a pipeline takes longer than expected or was launched against the wrong inputs, POST /v2beta/{+name}:cancel halts the underlying VM and stops billing. An agent watching cost or run time can call cancel and clean up associated GCS scratch space without manual intervention.

Example prompt: Call POST /v2beta/{+name}:cancel for an operation older than 12 hours and confirm the response is empty.

### Region-aware pipeline submission

Some workloads must run in specific regions for residency or pricing reasons. GET /v2beta/{+name}/locations enumerates supported regions so an agent can pick, for example, europe-west2 to keep patient-derived data in the EU before submitting the pipeline.

Example prompt: List locations under projects/{p} and pick europe-west2 as the parent for the next pipelines:run call.

### Agent-orchestrated genomics workflow via Jentic

An analysis agent receiving a sample manifest can build the right Pipeline definition, submit it through Jentic, and poll the operation until completion. Jentic surfaces all five Life Sciences operations so the agent does not need a Google Cloud client library to run a multi-step workflow.

Example prompt: Run google_lifesciences_run_pipeline with a Pipeline JSON, then poll until done=true and return metadata.events.

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | /v2beta/{+parent}/pipelines:run | Run a containerized pipeline |
| GET | /v2beta/{+name} | Get a long-running operation by resource name |
| GET | /v2beta/{+name}/operations | List operations within a parent location |
| POST | /v2beta/{+name}:cancel | Cancel a long-running pipeline operation |
| GET | /v2beta/{+name}/locations | List supported regional locations |

## Key resources

- **pipelines** — Submit container-based pipelines via the pipelines:run action
- **operations** — Track, get, list, and cancel long-running pipeline operations
- **locations** — List supported regional locations for the Life Sciences API

## Why Jentic

- **Setup:** Wiring the Cloud Life Sciences API by hand means configuring service-account OAuth2, narrowing to the cloud-platform scope, refreshing tokens, and learning its parent and operation resource paths yourself. Through Jentic you install once, import the Cloud Life Sciences API from the API Directory, store the credential once, and your agent calls it.
- **Permission scoping:** The parent and operation names travel in the URL path (/v2beta/{parent}/pipelines:run, /v2beta/{name}), so a rule can pin your agent to one project or location: it runs and tracks pipeline operations under that parent and nothing else. You choose the operations it may call, so if you grant only the run and get operations the agent can launch and monitor pipelines without cancelling them.
- **Credential handling:** Your Life Sciences service-account credential is stored once, encrypted, by your own Jentic One instance and injected at execution time as a short-lived cloud-platform access token. The JSON key never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'run a genomics pipeline' or 'cancel a long-running operation', and Jentic returns the matching v2beta operation with its parent and name templates so the agent calls the right endpoint without browsing the reference docs.

## Related APIs

- **Compute Engine API** — Compute Engine provisions individual VMs; Life Sciences orchestrates containerized pipelines on top
- **Cloud Build API** — Cloud Build runs container-based build steps; Life Sciences runs container-based scientific workloads
- **Cloud Logging API** — Cloud Logging captures stdout and stderr from pipeline actions

## FAQ

### What authentication does the Cloud Life Sciences API use?

OAuth 2.0 with the https://www.googleapis.com/auth/cloud-platform scope, typically using a Google Cloud service account that has lifesciences.workflows.run and operations.cancel IAM permissions. Through Jentic the service account JSON is encrypted in the vault and the agent receives only a short-lived access token.

### Is Cloud Life Sciences being deprecated?

Google announced sunset of Cloud Life Sciences with end of support on July 8, 2025. Existing v2beta workflows continue to run for now, and Google recommends migrating new workloads to Batch or GKE. This API is still useful for legacy genomics pipelines and as a reference.

### What are the rate limits for the Cloud Life Sciences API?

The default per-project quota is 1,000 pipelines:run requests per day and 60 operations.list requests per minute. Compute Engine quotas (vCPUs, persistent disk) bound how many pipelines can run concurrently.

### How do I run a pipeline through Jentic?

Run jentic search 'run a containerized pipeline on google cloud' to find POST /v2beta/{+parent}/pipelines:run, jentic load for the Pipeline schema, then jentic execute with a body containing actions, resources, and timeout. Jentic returns the long-running operation name for polling.

### How do I check if a pipeline has finished?

Poll GET /v2beta/{+name} with the operation name returned by pipelines:run. The response includes a done boolean - when true, either response or error is set, and the metadata.events array contains the timeline of pipeline steps.

### What does Cloud Life Sciences cost?

The API itself has no per-call charge. You pay for the underlying Compute Engine VMs (vCPU, memory, GPUs) and persistent disks the pipeline uses, billed at standard Compute Engine rates.

### Can I limit what my agent is allowed to do with the Cloud Life Sciences API?

Yes. Because you self-host Jentic One, your own rules decide which of the five Life Sciences operations the agent may call and which service-account credential it uses. Since the parent and operation names travel in the URL path, you can pin the agent to a single project or region so it only runs and tracks pipelines under that parent. Grant only the run and get operations and the agent can launch and monitor pipelines but cannot cancel them or list operations elsewhere.