BGSi Genome Storage Budget Calculator

Step 1

The genome collection

How many genomes we have and how fast the collection grows.

Short-read genomes today Short-read genomes added per year Long-read genomes today Long-read genomes added per year Years to plan for

Estimated data generated

Everything the collection produces by the end of each year, before any files are deleted or moved to cold storage. Uses the per-genome file sizes under Advanced assumptions in step 2, plus % overhead.

Step 2

AWS storage: where should each type of file live?

Each genome produces four kinds of files. Hot storage (S3 Standard) is always available but costs the most. Cold storage is 5–12× cheaper per month, but you pay a fee and wait hours each time files are brought back. Delete removes the file after processing. The same choices decide what goes on local disk (hot) and tape (cold) in step 3.

FASTQRaw sequencer outputBiggest files. Only needed if you want to redo alignment from scratch. CRAMAligned readsCompressed copy of the reads; enough to re-run most analyses. gVCFPer-genome variantsSmall. Needed when the whole cohort is re-analysed together. VCFFinal variant callsVery small. What hospitals and researchers actually use day to day.

Cold storage choice

Which cold storage? Full cold-archive restores per yearHow many times the whole cold archive is brought back

Advanced assumptionsFile sizes and AWS storage prices

File size per genome

FASTQ, short-read FASTQ / raw, long-read CRAM, short-read CRAM / BAM, long-read gVCF, short-read gVCF, long-read VCF, short-read VCF, long-read Index / QC files overhead Cohort joint VCFAlways hot · per 1,000 genomes

AWS S3 prices, Jakarta, per GB-month aws.amazon.com/s3/pricing

S3 Standard, first 50 TB S3 Standard, next 450 TB S3 Standard, over 500 TB Glacier Deep Archive Glacier Flexible Retrieval Glacier Instant Retrieval

Restore fees, per GB S3 pricing, "Requests & data retrievals"

Restore speed Deep Archive, standard Deep Archive, bulk Flexible, standard Flexible, bulk Instant Retrieval Restored data downloaded out of AWS Download out of AWS, per GBData transfer pricing

Prices from the AWS public price list for ap-southeast-3 (Jakarta), published 11–26 Sep 2026. Download tier shown is the >150 TB/month rate; the first 10 TB costs $0.132/GB. Not modelled: per-request fees (tiny at genome-file sizes) and early-deletion minimums (90 days for Instant/Flexible, 180 days for Deep Archive).

Step 3

Local storage pricing

What the same files would cost on our own hardware. Hot files go on a disk system, cold files go on LTO tape. The hardware is sized once for the data held in the final year and charged as a fixed yearly cost.

Disk, for hot files

Files you keep "hot" (always ready to open) sit on hard drives in our own server room.

Disk price per TBFull cost of the storage system per TB of raw disk: the hard drives plus the boxes and servers they sit in Extra disk for backup protectionRAID keeps data safe if a drive fails by storing extra copies or checksums. 25% means buy 25% more disk than the data Spare spaceDisks shouldn't run full. 20% means leave a fifth empty for growth and temporary files Fast working space (SSD)A small, fast drive pool used while files are being processed SSD price per TBCost of the fast working space, per TB

Tape, for cold files

Files you move to "cold" go on magnetic tape: very cheap per TB, but slower to read back. A robot library stores and loads the tapes.

Data per tapeHow much one tape cartridge holds. LTO-9, the current standard, holds 18 TB Price per tapeCost of one blank tape cartridge Copies of the dataHow many copies to keep. 2 = one in the library, one stored off-site in case of fire or flood Tape library (the machine)A cabinet with a robot arm that loads tapes into drives. This is its base price, before slots Price per tape slotEach tape in the library needs a slot. Bigger archives need more slots Tape drivesThe units that read and write tapes. More drives = faster to save or restore data Price per tape driveCost of one tape drive

Running costs

What it costs each year to own and run the equipment.

Hardware lifetimeHow many years the equipment lasts. Its price is spread evenly over these years Maintenance / supportYearly vendor support and repairs, as a % of the hardware price Staff to run itFull-time staff needed. 0.5 = half of one person's time Cost per staff memberYearly salary and overhead for one person

These defaults are budget placeholders, not vendor prices. Replace them with a quote from your storage vendor: LTO-9 tapes at $120, an enterprise library at $60,000 plus $100 per slot, 4 drives at $12,000, and disk at $150 per raw TB for a complete system. Not modelled: electricity and floor space for storage.

Results by year

Total storage budget over years, using everything entered above.

Cumulative storage spend

Your AWS plan, years

AWS, everything hot

Local disk + tape

Total data in year

Storage cost per year

AWS storage for each year is charged on the average amount of data held during that year, including restores. Local disk + tape is a fixed yearly cost.

Data held at the end of each year

Local disk + tape: yearly cost

AWS cost per file type over years

File	Plan	Size, yr	Your plan	If kept hot

Price sources

Item	Jakarta list price	Where to check

Checked 29 Sep 2026 against the AWS public price list (ap-southeast-3). Also useful: the AWS Pricing Calculator and S3 Glacier storage classes overview. Local disk and tape prices are placeholders; get a vendor quote.

Genome storage budget