Download every file your provider offers, check that each one is intact, keep at least three copies on two kinds of storage with one copy somewhere else, and encrypt anything that leaves your home. Prioritize the small, high-value files (the VCF and your report) first, then the large alignment and read files if you have room.
A cloud account or provider portal is not a backup. Portals change, accounts close, and you may not control how long data is kept.
What you are storing
A 30x whole genome produces several files at different levels of detail. Typical sizes from our FASTQ, BAM and VCF explainer, which your provider's files may differ from:
| File | What it is | Typical size | Keep it if |
|---|---|---|---|
| VCF | List of your variants | 200 MB to 1 GB | Always |
| BAM or CRAM | Reads aligned to a reference | BAM about 30 to 60 GB | You may want new analyses or structural-variant calling |
| FASTQ | The raw reads | About 50 to 100 GB compressed | You want the option to realign from scratch |
Storage prices change, so compare current prices before you buy drives or a cloud plan. A few hundred gigabytes is modest by today's standards, but it is more than most people casually back up.
Step 1: Download everything you can, soon
Retention rules differ between providers, and some portals offer files for a limited time or charge for later retrieval. Check your provider's terms, download all the files and reports as soon as they are available, and save the original filenames. Our list of questions to ask before buying whole genome sequencing includes asking how long files stay available.
Step 2: Check that the files are intact
Large downloads fail quietly. If the provider publishes checksums (MD5 or SHA-256 values), compare them with the ones your computer calculates. If not, run basic integrity tests: gzip -t file.vcf.gz checks a compressed VCF, and the free samtools suite includes a quickcheck command for BAM and CRAM files. A corrupt file discovered in five years is a bad day, so test now and write down the result.
Step 3: Consider shrinking the alignment file
CRAM is a format designed to give significantly better lossless compression than BAM. It does so by storing only how reads differ from a reference sequence, which means you must also keep the exact reference build used (see GRCh37 versus GRCh38). Convert only after checking the output, and never delete the original until you have verified the CRAM and backed it up.
Choose a test that hands you the files
A report you cannot export is not an archive. Dante Labs delivers your raw data, so you can store it, reanalyze it and move it to the tools you trust.
Compare Dante Labs WGS → Code GENOME — 10% offAffiliate link. Consumer sequencing is not a clinical diagnostic. Confirm medically important findings through your doctor.
If space is tight: keep the VCF and your original report in every location, keep one verified BAM or CRAM on at least two, and put the FASTQ files on the cheapest archive you trust. Re-run your integrity checks whenever you copy files to a new drive, because a copy that finished without an error message can still be damaged.
Step 4: Follow the 3-2-1 pattern
A common rule of thumb for any valuable data is 3-2-1:
- Three copies in total, counting the working copy.
- Two different kinds of storage, for example an external drive and a network or cloud service, so a single type of failure cannot take both.
- One copy off-site, away from fire, theft and floods at home.
Hard drives fail without warning, and one disk with the only copy is not a plan. Reconnect backup drives and test a restore once a year, and replace drives before they get old.
Open formats age better than portals. VCF, BAM and CRAM have public specifications, so software that reads them will probably outlast any single company's website. That is another reason to hold the files themselves instead of only a link to a report.
Step 5: Encrypt anything that leaves your control
A genome is identifying and cannot be changed if leaked. Before uploading to a cloud service or handing a drive to someone, encrypt it. Your operating system's disk encryption and common archive tools can do this, and a strong passphrase stored in a password manager is the key to everything, so keep a separate record of it. Avoid syncing raw genome files into a shared folder or a family photo library. Our DNA data privacy comparison explains why the data deserves this care, and can you really delete your DNA data covers what happens to copies held by companies.
When you share a file with a doctor, a counselor or an analysis service, send only what they need: usually the report or a single region, not the entire archive.
Step 6: Write a short README
Save a plain text note beside the files: provider, date received, test type, reference build, which file is which, and any checksums. Five years from now, "sample_final_v2.bam" will not explain itself. Leave out your name and anything that links the file to you if you share the folder.
Bottom line
Keep the VCF everywhere, keep the larger files if you can afford the space, verify every download, and encrypt copies that travel. The point of owning your raw data is to be able to reread it later, as we describe in why your genome needs reanalysis, and that only works if the files survive.
Common questions
How big is a whole genome file?
It depends on the file. A VCF of variants is typically hundreds of megabytes to about a gigabyte, aligned reads (BAM) tens of gigabytes, and raw FASTQ reads often more. Check your own provider's file list for exact sizes.
Should I keep my FASTQ files?
Keep them if you can afford the storage, because they let you realign to a newer reference build. If space is tight, a verified BAM or CRAM plus the VCF covers most future needs.
Is it safe to store my genome in the cloud?
Only with precautions. Encrypt the files before upload, use a strong unique passphrase, and understand the service's terms. Do not treat a single cloud account as your only copy.
What is CRAM and is it safe to use?
CRAM is a compressed alignment format that stores differences from a reference sequence. It is widely supported, but you must keep the matching reference build, and you should verify conversions before deleting the originals.
What if my provider deletes my data?
Download your own copy now. Retention rules vary, and once files are removed you may need to pay for retrieval or may be unable to recover them at all.
Sources and verification
File sizes repeat the typical ranges in our FASTQ, BAM and VCF guide and vary by provider and coverage. The 3-2-1 pattern is general data-protection practice and not specific to genomics. Check your provider's current retention terms.
Medical disclaimer. GenomeTesting.org publishes educational content about genetics and consumer genomic testing. Nothing here is medical advice, a diagnosis, or a substitute for care from a qualified clinician or genetic counsellor. Decisions about testing, treatment, or interpreting your own results should be made with a healthcare professional who can see your full history.
Affiliate disclosure. Some links on this page are affiliate links. If you buy through them we may earn a commission at no extra cost to you. This never changes which products we recommend or what we say about them.