How TileDB Scales Federated Queries with BeginNGS

Click here to watch the full Tech Talk

When we hear “rare disease,” it’s often easy to discount such ailments as outlier risks. But as discussed in our April Tech Talk, this thinking misses the real stakes of rare genetic conditions that affect 300 million people globally and are leading causes of child mortality in high-income countries.

“‘Rare disease’ is a bit of a misnomer,” said Jeremy Leipzig, Senior Product Manager at TileDB. “While each rare disease individually is rare, rare disease as a whole is not that uncommon—and these are manifest in children. However, there are many inherited diseases where we have actionable interventions.”

In this month’s webinar, we went deep on how TileDB powers the BeginNGS initiative, an ambitious effort to screen newborns globally for genetic disorders that are often actionable but too often go undetected. Key to the success of the BeginNGS mission is effective data sharing between hospitals that powers rapid diagnosis of rare diseases through federated queries.

How federated queries enable life-saving diagnoses for newborns

In newborn care, time is everything. “There’s a great chance to have a rapid diagnosis that can really affect the progression or outcome of a disease,” Jeremy said.

The BeginNGS program is expanding internationally to meet this challenge. Unlike static pipelines or regional dashboards, BeginNGS is built on a federated query architecture, which allows real-time querying across distributed datasets while respecting data sovereignty and patient privacy.

“We can't share patient identifiers or identifiable information, and we don't want to share static CSVs. What we really want are live queries of variants of interest in the block list that's growing on a weekly basis. The data privacy issue is very important, and that's why we have the Federated query.” - Jeremy Leipzig

Federated queries help solve the data sovereignty bottleneck

By comparing genetic information across institutions without transferring sensitive patient data, federated queries speed the diagnosis of rare diseases across the BeginNGS coalition while maintaining privacy. This breaks down the data sovereignty bottleneck in a trusted research environment.

This Tech Talk explored how TileDB’s federated query technology enables flexible and real-time genomic queries across institutions, hospitals, biobanks and research groups. Our federated system uses secure, pre-defined UDFs (user-defined functions) to execute code on remote datasets. For example, BeginNGS organizations are using TileDB to run federated queries that dynamically identify diplotypes (variant combinations likely to cause disease) across thousands of samples in less than 10 minutes.

This has the potential to revolutionize how we diagnose and treat rare disease in newborns. Here are some real-world outcomes enabled by this technology:

• 97% cost reduction over traditional raw VCF processing.

• Live dashboards and ad hoc queries delivered in seconds, not hours.

• Federated support for 342 curated genes, each tied to actionable interventions.

“In my experience, people have to work at a layer lower than where they’re really comfortable. So a scientist has to spend a lot of time doing data engineering. TileDB does a lot of that work for you so you can spend more of your time doing your actual role that you’re trained to do and less time messing around with infrastructure,” Jeremy said.

What’s next: TileDB Carrara

Jeremy also offered a glimpse of TileDB Carrara, our next-generation platform designed to simplify how researchers manage and share data post-primary analysis.

From secure workspaces to integrated UIs for non-technical users, Carrara is about turning TileDB from a high-performance backend into a user-friendly platform for discovery without giving up the raw power researchers love.

If you missed the April Tech Talk, or want to dig deeper into federated queries and BeginNGS, click here to watch the full Tech Talk.

Interested in partnering with BeginNGS? Let’s talk