Cloud germline variant calling

Submit FASTQ. Get a VCF back — plus a findings report you can actually read.

GPU-accelerated whole genome sequencing analysis, dispatched by API. Flat price per run. Every input, intermediate, and output file is deleted the moment your result is delivered.

New: every completed run now includes an informational findings report — real QC metrics and ClinVar-matched variants, automatically. See what's inside →

Run logLive
a3f9c1e2hg38Completed
7bd402a1hg38Running
e19f0c8dhg38Completed
44a7bb03hg38Completed
The pipeline
Input
FASTQ
→
Compute
RTX PRO 600096GB, dedicated per run
→
Pipeline
pbrun germlineNVIDIA Clara Parabricks
→
Output
VCF + Reportsigned link, refreshed on every view
New — included with every completed run, no extra charge

A VCF is a data dump. This is a starting point for reading it.

Every completed run now also delivers an informational findings report: real quality-control metrics computed from your actual alignment, and any exact matches against ClinVar's own published Pathogenic and Likely Pathogenic classifications — reproduced verbatim, never computed or inferred by us.

Included today
✓ Mean depth, mapping rate, % bases covered — from the real alignment, not estimated
✓ Exact-match comparison against ClinVar Pathogenic / Likely Pathogenic entries
✓ Gene, condition, zygosity, and a direct ClinVar link for every match
✓ One signed PDF, delivered alongside your VCF, same job
Not yet included
– Pharmacogenomics / drug-response guidance
– Full ACMG secondary-findings gene panel
– Carrier screening, ancestry, polygenic risk scores
– Variants of uncertain significance — not reported, by design

Most runs will show no primary findings — that's the expected, common result, not a failure of the test. For the majority of samples ClinVar simply has no matching Pathogenic or Likely Pathogenic entry, and the report says so explicitly rather than leaving the page blank.

Spec sheet

What you're actually paying for.

Reference buildhg38
ComputeDedicated GPU per run — not a shared queue
Billing modelFlat price, known before you submit — not metered by runtime
Pay-as-you-go$30 per run
Run bundle$2,000 for 100 runs — $20/run
Data retentionNone — deleted on delivery, every run
DeliverySigned download link, expires within hours
AccessAPI — submission and status, built for automation
Data policy

Your samples don't become our liability.

Nothing persists after delivery
FASTQ input, intermediate BAM files, and the final VCF are all removed from every system component immediately after your result is delivered.
No shared storage
Each run's data exists only for the duration of that run, on infrastructure provisioned for it and torn down after.
Metadata only, permanently
What we keep on record is limited to job status, timestamps, and cost — never sequence data, never variant calls.
Data & compliance

Every layer that touches your data runs on certified infrastructure.

GPU compute executes exclusively on our infrastructure provider's Secure Cloud tier — independently certified to SOC 2 Type II, SOC 3, HIPAA, and GDPR standards, with no exception.

✓SOC 2 Type II
✓SOC 3
✓HIPAA
✓GDPR

In-transit storage, while a run's files exist briefly before delivery and deletion, runs on Railway's managed Buckets — also SOC 2 Type II and SOC 3 certified, with a HIPAA BAA and GDPR DPA available on request.

✓SOC 2 Type II
✓SOC 3
✓HIPAA (BAA on request)
✓GDPR (DPA on request)

Need a signed Business Associate Agreement or Data Processing Agreement for your own compliance requirements? Contact us — available for qualifying accounts.

Cost comparison

Cost per run, against owning the hardware yourself.

What a lab running the same GPU class on its own physical server would spend per run, once hardware, staff, and facility costs are spread across actual usage.

What's included in the self-hosted estimate
Server hardware, amortized (same GPU class)$600 / mo
DevOps / bioinformatics support$3,000 / mo
Power, cooling, facility$300 / mo
GONZA-K, per run $30
Self-hosted, per run at this volume $120
Estimated savings per run $90
Self-hosted fixed costs are spread across your monthly run volume — the more you run, the lower this per-run figure gets. Actual costs vary by lab.
Technical overview

Get the pipeline documentation.

Turnaround benchmarks, the exact pbrun invocation, and what's covered under the flat price — sent as a PDF.

>