How to transfer large FASTQ & BAM files

A single lane of sequencing can produce tens of gigabytes of FASTQ; an aligned BAM or a single-cell matrix isn't far behind. Then a collaborator says "just send it over" — and the trouble begins. This guide covers why normal sharing breaks on genomic data, how the classic command-line options really compare, and the fastest secure way to move multi-gigabyte files between labs.

Why sequencing files break normal sharing

The formats researchers exchange every day are simply too big for consumer tools:

• FASTQ — raw reads, routinely 5–50 GB per sample (more for deep WGS).
• BAM / CRAM — aligned reads; whole-genome BAMs often run 50–150 GB.
• scRNA-seq matrices & 10x outputs — many files, easily tens of gigabytes zipped.

Email caps out around 25 MB. Consumer cloud drives technically hold the file, but choke on multi-gigabyte uploads over a lab connection, silently pause, or wrap everything in a login wall the recipient doesn't have. And genomic data from humans carries HIPAA / GDPR obligations that a shared Drive folder wasn't built for.

The options, compared honestly

Every method below works — the question is what it costs you in setup, size limits, and hassle for the person on the other end.

Command-line tools like scp, rsync and Globus are excellent inside an institution — but they assume both ends have accounts, keys, or matching endpoints. For a quick hand-off to a collaborator at another lab (or a reviewer, or a clinician), that setup is the bottleneck.

The fastest path: send it from the browser

A browser-based transfer avoids the whole setup problem. With BioTransfer the flow is:

1. Drag the file in. It's sliced into chunks and streamed directly to object storage — so the ceiling is your connection, not an inbox limit. Nothing is capped at a few gigabytes.
2. Integrity is checked automatically. A checksum is computed so a truncated FASTQ never slips through to break alignment downstream.
3. Add biology-friendly metadata (organism, assay) so the recipient knows what they're getting.
4. Share one link. The recipient downloads straight from storage — no account, no login.

Best practices for moving sequencing data

• Verify a checksum (MD5/SparkMD5) on both ends. Silent corruption in transit is the classic cause of "the BAM won't index."
• Compress first. FASTQ should travel .gz; use CRAM over BAM where your pipeline allows — often 40–60% smaller.
• Send metadata with the data — reference genome, assay, sample sheet — so provenance survives the hand-off.
• Set an expiry and don't leave raw human reads sitting in a shared folder longer than you need.
• Never put identifiers in the filename of a link you share publicly.

Looking for the data itself first? Find and download public datasets with the GEO Dataset Finder, or explore more free research tools on the BioTransfer home page.