The instrument produces faster than the network drains
A sequencer writes tens to hundreds of gigabytes of FASTQ per run, and whole-genome runs push into the terabytes once BAM and CRAM files are added. A cryo-EM or light-sheet session produces terabytes in a night. That data lands on an acquisition PC, often on an isolated VLAN, and has to reach a core facility, several labs, a bioinformatics team and collaborators abroad.
The usual answers are a shared drive that copies at TCP speed over distance, a per-GB transfer service, or a portable drive. None tells you whether the copy is bit-for-bit identical, which matters when the run will be reanalysed for years.
What Boita does about it
Boita moves data with an accelerated transfer engine over UDP with its own rate control. A transfer climbs to your line rate and holds it whether the collaborator is on campus or on another continent, and large runs go as parallel sessions sized to your plan. Transfers checkpoint as they run, so a dropped link or a rebooted server resumes rather than restarts, and completed files are never re-sent.
The server records a checksum for every file. Choose SHA-256 or SHA-512 for the workspace, compare it with the hash your pipeline wrote, and the copy is verified rather than assumed. A run folder of thousands of files moves as one job with its structure intact.
Fits the instrument PC, the cluster and the pipeline
On the acquisition PC, the desktop app for macOS, Windows or Linux watches the run output folder and uploads each file once it has stopped changing, so a BAM still being written never ships. On a headless acquisition server or a cluster login node, the Linux agent runs as a systemd service with watch folders up and down. Allowed hours and bandwidth windows are enforced by the transfer server, so a run drains overnight.
The boita command line drops into pipeline scripts, and the REST API with a Python SDK and signed webhooks (Pro and above) lets a workflow start analysis the moment a run has landed. A core facility gives each lab an upload portal with a form for sample identifiers, so data comes in labelled, in its own dated inbox folder.
A worked example
A 500 GB whole-genome run leaves a core facility on Pro, over a 1 Gbps campus line, in about two and a half hours; the plan ceiling of 450 Mbps is the limit, not the line. A 3 TB cryo-EM session on Studio over the same line is line-limited and takes about six and three-quarter hours, an overnight window. What a collaborator abroad sees on the way down depends on distance and loss on their side; the calculator on the home page models it.
- Line-rate, resumable transfers of FASTQ, BAM, CRAM and microscopy stacks
- Watch folders on the instrument PC; the Linux agent on acquisition servers
- boita CLI, REST API and Python SDK for pipelines (Pro and above)
- Upload portals with forms for labs sending samples in
- Checksum on every file; receipts with a verification report
Proof and controls
Delivery receipts list every file, its size, its checksum, the timestamp and who downloaded it, with a CSV verification report and a PDF on Studio and above: what a sponsor asks for. Share links carry an expiry, an optional password and a download limit; recipient verification by email code names the person who downloaded. Sender-side encrypted delivery with a passphrase keeps a dataset unreadable to anyone, including Boita, until the recipient enters it. SSO, retention rules and legal hold are on Enterprise.