Singular Genomics Scales On-Prem NetApp Sequencing Deliveries 20x with Files.com

Singular Genomics builds the G4X, a next-generation sequencing instrument that pairs sequencing with an astronomy-grade camera to produce spatial images of tissue at subcellular resolution. Academic medical centers, cancer research programs, and pharmaceutical companies use it two ways. Some run G4X instruments in their own labs, and Singular has a fleet of them deployed at client sites. Others send Singular their samples: the company's services lab sequences and analyzes biopsy and tissue samples and returns the results as datasets.

Either way, what the client is waiting for is data, and the quantities are enormous. A single sequencing run can generate roughly 100 TB of raw data. The customer-facing output alone runs 1 to 8 TB per job, and a full client delivery runs 5 to 20 TB. All of it is born on premises: sequencers write to Singular's NetApp storage. Handing a multi-terabyte dataset to an outside institution is not a side task at Singular. It is the last step of the client service.

With Files.com, that delivery workflow absorbed roughly 20x growth in nine months without added transfer engineering, leaving the same six-person IT and InfoSec team to support the entire company’s infrastructure.

Every Delivery Was an Improvisation

The path from sequencer to client had grown up ad hoc. The NetApp systems held sequencer output in deeply nested, inconsistent folder structures, and users mounted those directories directly on their own Linux workstations. Sharing anything externally meant a manual upload, every time. Deliveries moved by hand over SMB. At least one client was told to buy their own Synology NAS so a Singular staffer could remote into it and push the data across. Other datasets were parked in AWS and left there, with Singular paying for storage month after month and no way to know whether the client had ever collected the files.

The cost landed on both sides of the handover. Internally, the six-person team was hand-carrying terabytes as a routine part of client service. Externally, the institution waiting for sequencing results depended on whatever transfer method someone could improvise, and nobody could confirm what had actually arrived. For genomic data, the loose ends were also a liability: datasets with no owner and no expiration date, accumulating in cloud storage.

The Obvious Fix Had Already Been Tried

A managed SFTP service is the standard answer, and Singular had already attempted one. A build on AWS Transfer Family went nowhere. The team found it complex and convoluted to operate, and the setup it left behind was undocumented, with no one who knew how to run it.

The harder constraint was where the data lives. Sequencer output is born on on-prem NetApp, reached through two separate Active Directory domains. Any cloud transfer endpoint meant re-uploading terabytes before each delivery, which doubled the work the fix was supposed to remove. And the workload kept growing: every G4X in the field and every new services client added deliveries to a process running on individual effort.

What the fix had to do was specific. Reach the NetApp systems where the data lands, without copying everything to the cloud first. Give each client institution a governed account, a notification when files arrive, and a way to verify the transfer. Delete sensitive data on a schedule instead of letting it accumulate. And carry the compliance paper that genomic data demands: a HIPAA BAA and a GDPR DPA. Singular selected Files.com to be that delivery layer.

An Agent on the Storage Where the Data Is Born

Files.com became the delivery layer between the sequencer estate and every client institution, sitting in front of the NetApp rather than replacing it.

Singular installed the Files.com Agent inside its network, and Files.com Remote Server Mounts presented the sequencer-output directories on the NetApp systems as ordinary folders on its Files.com site. The Agent connects outbound only, so nothing on the sequencing network is exposed to the internet. A dataset sitting on the NetApp is already on the delivery platform. Nobody uploads terabytes to a transfer server anymore; the transfer platform reads the storage the sequencers wrote to.

Every delivery starts with a short intake, and one question in it decides the account's whole lifecycle: is this a one-time delivery or an ongoing relationship? One-time clients get an account that is deleted when the delivery completes. Ongoing clients keep a standing account, so the next dataset needs no setup at all. When files land, Files.com sends the client an automated notification, and the client downloads over SFTP or in the browser, on Singular's own branded domain. MD5 checksums exported against the SFTP logs confirm the client received the exact files, which closes the question the old process could never answer: did the data actually arrive?

Governance runs on its own clock. Client-facing files expire 30 days after upload under a Files.com retention policy, so delivered datasets clean themselves up instead of accruing cost and risk. Internal access runs through Azure SSO with two-factor authentication enforced, and the activity log backs the quarterly access audits the team runs.

From 2.8 TB a Month to More Than 50

With the Files.com workflow in production, Singular replaced a per-client improvisation with one automated delivery path for the entire deployed G4X fleet.

“It's pretty much automated once we get the data into the system.”

Volume through the workflow grew roughly 20x in nine months, from about 2.8 TB a month at go-live to more than 50 TB stored and transferred in a single month, absorbed on the same workflow with no transfer engineering added. Institutional-scale handovers became routine, with multiple client deliveries in a single month.

Delivered files delete themselves after 30 days. No genomic dataset sits in storage indefinitely while the company pays for it and wonders whether the client ever picked it up.

Adding the next client is an intake decision and an account. Adding the next instrument to the fleet adds volume, not process, so delivery capacity scales with the fleet rather than with the six-person team's hours.

The Handover Became the Dependable Step

Today, the last step of Singular's service no longer depends on who is available to carry the data. A run finishes, the output lands on the NetApp, and the client's account and notification are already in place, with the 30-day clock starting at upload. The team that once remoted into client-bought NAS hardware to push terabytes across the internet now watches deliveries go out, get confirmed, and expire without touching them.

Singular never had to move its data to fix its delivery. The sequencer output stayed on the NetApp where it was born, and Files.com mounted that storage and ran the delivery on top of it. For a company whose product is a multi-terabyte dataset, that turned the handover, once the most improvised step in the operation, into the most dependable one.