We provide installation and run instructions for both local runs and cloud runs with Seqera Platform.
Local installation is best for development, testing, or running on a single server. We recommend you perform a local installation regardless of whether you will eventually also run at scale in Seqera Platform so that you can simply run minimal test pipelines.
Cloud execution with Seqera Platform is recommended for production runs at scale. Seqera Platform (formerly Nextflow Tower) provides a web interface for launching, monitoring, and managing Nextflow pipelines on cloud infrastructure.
Local Installation of Nextflow¶
Local Prerequisites¶
Java: Version 11 or later
Nextflow: Version 25.04.0 or later
Docker: Engine must be installed and running
Local Setup¶
Install Nextflow:
curl -s https://get.nextflow.io | bash mv nextflow /usr/local/bin/Verify installation:
nextflow run broadinstitute/nf-pooled-cellpainting -profile docker --help -r dev
After successfully verifying installation, we recommend you run the local test example.
Cloud Setup of Seqera Platform¶
AWS Prerequisites¶
Seqera Platform Account: Access to a workspace at cloud.seqera.io
AWS Account: With permissions to create Batch resources
AWS Credentials: Configured in Seqera Platform
S3 Bucket: For work directory and data storage
Setting Up an AWS Batch Compute Environment¶
From your Seqera Platform workspace, navigate to Compute Environments → Add Compute Environment.
Basic Configuration¶
| Setting | Recommended Value | Notes |
|---|---|---|
| Name | AWSBatch_pooled_cellpainting | Descriptive name |
| Platform | AWS Batch | |
| Credentials | Your AWS credentials | Must have AWS Batch permissions |
| Region | us-east-1 (or your preferred region) | Should match your S3 bucket |
| Work directory | s3://your-bucket/work | Pipeline scratch data |
Seqera Features (Optional but Recommended)¶
| Feature | Description | Recommendation |
|---|---|---|
| Wave containers | Container provisioning service | Enable for easier container management |
| Fusion v2 | Virtual distributed file system for S3 | Enable for faster S3 access |
| Fast instance storage | NVMe for faster I/O | Enable if using Fusion v2 |
| Fusion Snapshots | Auto-restore on spot interruption | Enable for spot instance resilience |
Config Mode: Batch Forge¶
Select Batch Forge for automated queue creation. Seqera will create:
A head queue (for the Nextflow process)
A compute queue (for pipeline tasks)
Forge Configuration¶
| Setting | Recommended Value | Notes |
|---|---|---|
| Provisioning model | Spot | Cost-effective; use On-demand for critical runs |
| Max CPUs | 2000 | Total CPU pool; adjust based on workload |
| Allowed S3 buckets | Your data bucket(s) | Grant read-write access |
| EFS/FSx | None | Not required for most use cases |
Advanced Options: Instance Types¶
For image processing workloads, select instance families with:
Good compute-to-memory ratio
NVMe storage (if using Fusion)
Recommended instance types:
c6id- Compute-optimized with NVMem6id- General-purpose with NVMer6id- Memory-optimized with NVMe
After setup, we recommend you run the test example.