GENI Workflows

Nextflow · Cromwell · AWS Batch · GCP Cloud Batch

Your pipelines. Your cloud account.Your bill, itemised.

GENI Workflows is a control plane for genomics pipelines. You get one API and one CLI across every account and both clouds; the sequencing data and the compute never leave the accounts you already own.

Read the docsnpm install -g geni-bioinfo
wgs-germlinev2.1.0
Bioinfo Prodspot-m5us-east-1
FASTQC
BWA_MEM (1/8)
BWA_MEM (2/8)
BWA_MEM (3/8)spot reclaimed → fallback queue
BWA_MEM (4/8)
MARK_DUPLICATES
BQSR
HAPLOTYPECALLER (1/4)
HAPLOTYPECALLER (2/4)
GENOTYPE_GVCF
214 tasks6h 12m wall1 spot reclaim$9.84 billed
  • succeeded
  • running
  • queued
  • reclaimed

Cloud genomics gets expensive in two directions

Running Nextflow or Cromwell on cloud infrastructure yourself means owning Batch compute environments, IAM, image registries, spot interruptions and retry logic — per account, per region, forever. Every new client account starts that work again.

Managed services remove that work and charge for it. And when the invoice arrives at the end of the month, it is one number for the whole account: no answer to which workflow, which sample or which client it came from.

What you manage

Six objects, both clouds

The same six resources describe an AWS account and a GCP project. Learn them once.

Environment
An AWS account or GCP project, plus a region. Register as many as you run in — dev, prod, one per client.
Engine
Runs Nextflow or Cromwell for each submission, scoped to the exact buckets it is allowed to read and write.
Queue
Where tasks execute, on AWS Batch or GCP Cloud Batch. Keep a spot queue and an on-demand queue; pick one per run.
Storage
S3 and GCS buckets, whether GENI created them or they already existed in your account.
Registry
Your ECR repositories. Engines in an environment can pull the images you built for your own tools.
Plugins
Optional per-environment extras — a Sentieon license server, an SNS topic for submission status changes.

Cost

Every run has a price on it

GENI reconciles against the AWS Cost and Usage Report and the GCP FOCUS billing export — actual billed cost, not an estimate from a price list. Attribute it by workflow, environment, queue, instance type, or by the tags you put on a submission.

How tagging works
Cost by tag · last 30 days
clientrunsbilled
acme128$1,204.60
hospital-sp74$688.12
research-x31$294.05
validation12$71.88

Start with the CLI

Install it, activate the account your admin created, and list what is already running.

npm install -g geni-bioinfo
geni auth activate --token <token from email>
geni submission list
Getting started