When the infrastructure matches the science

In CIQ’s recent blog post marking ten years of Apptainer, CIQ traced how Singularity gave research scientists something they had never had before: a software environment that kept pace with the science. Version conflicts disappeared. Reproducibility became achievable. A decade of adoption on some of the world's fastest machines proved the idea out.

That post ended with a forward-looking observation. Containers solved portability. Portability was never the whole problem.

FYR Bio is a good illustration of what comes next.

What FYR is building

FYR is a precision medicine company based in Missoula, Montana. Their platform, EVO, enriches extracellular vesicles from a standard blood draw and applies AI-enabled multi-omic analysis to generate protein and RNA biomarker data across neurology and oncology. Their work spans neurology and oncology, and includes an active collaboration with Mayo Clinic and a National Cancer Institute-funded role in a Phase IIb glioblastoma trial.

The output of FYR's platform is the critical data that pharmaceutical partners use to stratify patients, monitor treatment response, and design more targeted trials. Producing that data at the speed and reproducibility pharma partners require puts real pressure on the infrastructure underneath it.

FYR runs eight to ten distinct multi-omic pipeline types in parallel. Each pipeline handles protein profiling, RNA sequencing, and AI-driven analysis across longitudinal clinical samples. For pharma partners who evaluate data quality as a direct proxy for credibility, the results have to be consistent, reproducible, and delivered at the pace the market requires.

The infrastructure problem

Before Fuzzball, the compute environment underneath FYR's science created friction at every layer.

New team members needed days of onboarding per pipeline. Results varied across users and environments in ways that required ongoing correction. Scientists were managing compute resources instead of running analyses. And as FYR's pipeline portfolio grew and pharma partnerships deepened, the gap between what the science required and what the infrastructure could deliver became harder to close.

"Even in highly optimized wet labs, physical processes have limits on how much they can be accelerated. When working with highly complex data, we often gain more leverage on the computational side."
— Claire Seibold, Director of Software and Data Analytics at FYR

The computational side had to become a strength.

Download the FYR Case Study →

The Fuzzball solution

Fuzzball is the HPC workflow orchestration platform built by the same engineers who created Apptainer. It is the production layer above the container: the system that takes a containerized environment and makes it schedulable, reproducible, and scalable across on-premises clusters, cloud resources, and national computing facilities.

Fuzzball's Workflow Catalog gives research teams a shared library of validated, containerized pipelines. Every workflow job runs in a container by default, so results are consistent across users, compute environments, and the full path from development to production. New team members run pipelines immediately. Senior researchers stop serving as onboarding support. With consistent results guaranteed by design, FYR’s scientists can focus on analyses. The infrastructure handles scheduling, consistency, and environment management.

For teams with existing scheduler investments, Fuzzball integrates with PBS and Slurm. It adds the orchestration layer without requiring a wholesale replacement of existing infrastructure.

Download the Fuzzball Solution Brief →

What changed for FYR

With Fuzzball, FYR achieved 100x throughput potential across their multi-omic pipeline suite. The pipelines did not change. The science did not change. The infrastructure underneath them finally matched what the science needed.

New team member onboarding dropped from days to minutes, per pipeline. Across all eight to ten pipeline types, results are now consistent regardless of which researcher runs the job or the compute environment in which it runs. FYR's pharma partners receive data that is reproducible by design.

"Fuzzball gives us the ability to implement and scale these pipelines for 100x the throughput in a fraction of the time," said Seibold. "It delivers results to our pharma partners at the speed and scale that today's innovation-driven discovery market demands."

FYR's value to pharmaceutical partners is the quality and speed of its biomarker data. Partnerships in precision medicine are built on the confidence that results are reproducible across time points, across clinical sites, and across longitudinal samples collected over months or years. Fuzzball makes that confidence a structural property of FYR's platform. Reproducibility is no longer managed manually but is guaranteed by design.

Download the FYR Case Study →

From portability to production

FYR dynamically observes disease and treatment response to produce better biomarkers, better trials, better drugs, and ultimately better outcomes for patients. Progress toward that mission requires computational infrastructure that can run at the pace the science demands and scale as the program portfolio grows.

FYR's pipeline infrastructure scales with their program portfolio. Each new pharma partner and each new disease area runs on the same Fuzzball-orchestrated foundation, inheriting the same reproducibility guarantees from the first run.

FYR's science team scales into new disease areas and pharma partnerships. Fuzzball manages the orchestration.

The research community spent a decade building workflows on Apptainer because containers solved a real problem. The next problem in the stack is orchestration: taking those containerized environments and making them operational workflows at scale, across real-world compute resources, for research teams who should be focused on the science and not the scheduler.

FYR Bio shows what it looks like when the infrastructure matches the science.

Download the FYR Bio Case Study →
Download the Fuzzball Solution Brief →

---

CIQ is the founding support and services partner of Rocky Linux and the creator of Fuzzball, the AI and HPC workflow orchestration platform built for research and enterprise computing at scale. Learn more at ciq.com.