☰

Cyclomics

From research kit to IVDR-certified diagnostic

Industry
Molecular Diagnostics
Headquarters
Utrecht, Netherlands
Public information as of
February 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of Cyclomics's published strategy and is not endorsed by, or produced in cooperation with, Cyclomics. Company website

Strategic priorities

Cyclomics originated as a spin-out from University Medical Center Utrecht and is preparing to move its commercial laboratories to the Matrix ONE building at Amsterdam Science Park during 2026. The company's core technology, CyclomicsSeq, uses circularization and rolling-circle amplification to read circulating tumor DNA from a single blood sample, and pairs it with the Epinn AI framework, a Software as a Medical Device (SaMD) classifier that returns tumor typing in around 90 minutes.

Capital structure shifted decisively in January 2026, when Cyclomics closed a €2.6 million round led by Myosotis Investments and the Oncode Bridge Fund and was acquired by NASDAQ-listed Lunai Bioworks. The combined event funds the team expansion explicitly needed to get IVDR-marked products ready for use in clinical centers, and brings the operational and cybersecurity expectations of a public-market parent onto a research-stage organization.

Three data streams come out of every CyclomicsSeq run: sequence, methylation and structural variation. Today they are handled separately, and the Epinn classifier trains on each in turn. The work that connects them — ontology, ingestion and retrieval — sits in front of any claim that liquid-biopsy monitoring is a routine clinical tool, and that is where the immediate digital priorities concentrate.

Cyclomics also runs an Expertise as a Service consultancy that provides genomics work to external life sciences partners; that branch is a way to bring in revenue while the IVDR transition completes, and it draws on the same wet-lab and dry-lab capability that the diagnostic kits will scale.

Challenges we see

  • Operations Manufacturing

    Scaling library preparation without losing turnaround time

    CyclomicsSeq library preparation combines circularization and rolling-circle amplification and is currently performed by skilled scientists. Cyclomics has guided to a three-day turnaround time from blood draw to result, and the company is preparing to scale to hospital-level distribution.

    Where every library is prepared by hand at the bench, the throughput ceiling sits on the available scientist-hours rather than on the chemistry. Standing the workflow up as a programmed procedure widens the ceiling while leaving the analytical chemistry itself untouched.

  • Compliance Regulatory

    Producing IVDR-grade evidence from research-stage pipelines

    The European IVDR (In Vitro Diagnostic Regulation, 2017/746) significantly raises the bar for clinical validation and technical documentation. Cyclomics has publicly named the IVDR transition as the use of proceeds for the January 2026 round, and Epinn falls under SaMD (Software as a Medical Device) regulation, which adds a validated software development lifecycle to the workload.

    Evidence that is acceptable for a research publication is not, on its own, acceptable for a notified body. Re-running the same analyses against IVDR documentation requirements is work the team can predict and schedule once the gap between research output and dossier output is mapped.

  • Digital Integration

    Closing the gap between sequencing speed and processing speed

    Cyclomics's acceptance into the Google for Startups Cloud Program was described as driven by 'growing computational needs for big data and AI model training,' which reads as previous on-premise or hybrid infrastructure being outgrown by NanoRCS data volumes.

    When data generation outpaces data processing, latency shows up first in the queues behind the sequencers. A pipeline architecture that absorbs bursty NanoRCS loads as a designed property keeps turnarounds predictable as sample volume rises.

  • Digital Operations

    Bringing IT and OT onto the same network at Matrix ONE

    The relocation to the Matrix ONE building at Amsterdam Science Park requires physically moving Oxford Nanopore GridION and MinION devices and re-validating them in a new network environment. The site is a greenfield for IT/OT (Information Technology / Operational Technology) design, since the previous Utrecht-based research footprint is not being copied over.

    The first time equipment sits on a new network is when its data path gets defined. Agreeing the protocol and segmentation standards before commissioning determines whether device data is reachable from the rest of the stack or has to be excavated later.

  • Digital Integration

    Reading sequence, methylation and structural data as one result

    The single-pot multi-omics patent produces three data types from each liquid biopsy sample. Cyclomics runs clinical evaluations with UMC Utrecht and Erasmus MC, which means the same three-stream shape has to be aligned across multiple study sites before training the Epinn classifier.

    Where the three streams live in three schemas, the classifier sees them as three separate evidence bases. Carrying them into one semantic layer lets the model weigh sequence, methylation and structural evidence together rather than reconciling them per report.

Opportunities, by urgency and business impact

Each bubble is one opportunity, numbered to match the list below. Further right means it bites sooner; higher means a bigger effect on the business. A bigger bubble means a bigger implementation effort.

Source: A4BEE analysis of public sources

  1. Automating CyclomicsSeq library preparation

    CyclomicsSeq library preparation combines circularization and rolling-circle amplification and is currently a bench skill held by scientists. The throughput ceiling on commercial kit production sits on the available scientist-hours, and any scale-up has to bring that ceiling up.

    Programming the circularization, amplification and quality-check steps onto a workstation turns library prep from a per-sample manual operation into a per-batch procedure. The chemistry is unchanged, and the scientist-hours that came out of pipetting move into supervising the run and triaging exceptions.

    • Cyclomics 10_Analysis, January 2026
    • Cyclomics_DeepResearch, Strategic Intelligence Report

  2. Building an IVDR-ready documentation layer for the SaMD pipeline

    Cyclomics is moving from Research Use Only to IVDR-certified diagnostic kits, and Epinn is a Software as a Medical Device. The IVDR dossier, the SaMD software development lifecycle and the clinical evidence base have to be assembled from the same underlying records.

    Treating the IVDR and SaMD evidence as a structured output of the same pipelines that produce the science means the dossier work does not start from a manual re-collection every cycle, and every change to a method leaves a trace that a notified body can follow.

    • Cyclomics 10_Analysis, January 2026
    • Cyclomics_DeepResearch, Friction Matrix, IVDR Enforcement

  3. Moving NanoRCS bioinformatics onto a resilient cloud architecture

    Cyclomics's previous infrastructure was described as outgrown by NanoRCS data volume, and the company was accepted into the Google for Startups Cloud Program on that basis. Bioinformatics runs currently share the same failure modes as the underlying on-premise or hybrid stack.

    Running the base-calling and consensus-calling pipelines on a high-availability container architecture on Google Cloud decouples pipeline uptime from any single host and lets capacity scale with sequencing volume rather than with the next hardware purchase.

    • Cyclomics_DeepResearch, Digital Friction: Bioinformatics Pipeline Bottlenecks
    • Cyclomics 10_Analysis, Cloud Bioinformatics Pipeline Optimization

  4. Designing the IT/OT architecture for Matrix ONE before commissioning

    Matrix ONE is a greenfield facility. Oxford Nanopore GridION and MinION devices are being moved in and re-validated, and the network they land on will determine whether device data is reachable from the rest of the stack from day one.

    Setting the protocol standards, segmentation zones and equipment data requirements before the instruments are commissioned puts interoperability into the design rather than into a later integration project, and leaves a clean reference for the next site Cyclomics opens.

    • Cyclomics_DeepResearch, Operational Friction: Infrastructure Transition Risks
    • Cyclomics 10_Analysis, Matrix ONE Facility Digitalization

  5. Unifying sequence, methylation and structural data behind one ontology

    Each CyclomicsSeq sample produces three distinct data streams — sequence, methylation and structural variation — and they are handled separately today. Multi-site clinical evaluations (UMC Utrecht, Erasmus MC) multiply the alignment problem.

    Defining an explicit ontology for sample, assay, run, methylation call and structural variant lets all three streams load against one model. The Epinn classifier can then train and query across the full evidence base instead of reconciling three tables per sample.

    • Cyclomics_DeepResearch, Digital Friction: Data Silos in Multi-Omics
    • Cyclomics 10_Analysis, Unified Multi-Omics Data Platform

What we'd propose

  • Digital Lab

    Automated library preparation workstation for CyclomicsSeq

    We design and integrate a workstation that performs CyclomicsSeq circularization, rolling-circle amplification and inline quality checks, replacing per-sample manual pipetting with a programmed per-batch procedure that the existing scientists supervise.

    • Workstation integration for library prep

      From bench to programmed procedure

      Integrate a liquid-handling workstation into the CyclomicsSeq workflow so that circularization and rolling-circle amplification steps are executed by the system with the same reagent volumes and timings as the current manual protocol. Method parameters live in versioned files rather than in technician memory.

    • Barcode-driven batch tracking

      Every sample traceable end-to-end

      Read sample and reagent barcodes at intake and at each step, so that every library in a run is tied back to its source specimen, its method version and the operator or system that handled it. The chain of custody is generated by the workflow rather than reconstructed for an audit.

    • Inline quality check before sequencing

      QC at the workstation, not after

      Place an inline QC station between library prep and sequencing that reads library concentration and fragment size, with pass/fail criteria that gate the run automatically. QC results reach the LIMS (Laboratory Information Management System) with the instrument identity and timestamp attached.

    • Throughput ceiling moves from the available scientist-hours to the workstation schedule.
    • Library-to-result variability comes from the method file rather than from the bench.
    • Scientists' time goes into supervising runs and triaging exceptions instead of pipetting.

    We've done this before

    See how we do Digital Lab
  • Digital Lab

    Building an IVDR-ready documentation layer for the SaMD pipeline

    We structure the pipelines that produce CyclomicsSeq results and Epinn classifications so that the same records feed the IVDR dossier and the SaMD software development lifecycle, with an audit trail a notified body can follow from method to release.

    • ALCOA+ data capture

      Records that hold up under inspection

      Capture every data operation against ALCOA+ principles — Attributable, Legible, Contemporaneous, Original, Accurate — so that each value carries the user or system identity, the method version and the timestamp that prove when and how it was produced.

    • Validated software lifecycle for Epinn

      SaMD evidence from the same pipeline

      Run the Epinn classifier's training, evaluation and release steps through a validated software development lifecycle that meets IEC 62304 (the international standard for medical device software lifecycle processes), so the SaMD dossier and the IVDR dossier are produced from the same auditable record.

    • Electronic batch records for kit production

      Digital records of every lot

      Replace paper-based kit manufacturing records with electronic batch records that capture reagents, equipment, environmental conditions and operator or system actions for every lot, with deviation handling built into the record rather than tacked on after release.

    • The IVDR and SaMD dossiers are produced from the same records as the science, not rebuilt separately.
    • Every method change leaves a trace that a notified body can follow to its evidence.
    • Deviations are handled in the record at the moment they occur, not reconstructed later.

    We've done this before

    See how we do Digital Lab
  • Enterprise AI

    High-availability bioinformatics on Google Cloud

    We design and deliver a containerized, high-availability architecture on Google Cloud for the NanoRCS base-calling and consensus-calling pipelines, so that pipeline uptime and capacity scale with sequencing volume rather than with the next hardware purchase.

    • Containerized pipeline deployment

      Pipelines that move between hosts

      Package the NanoRCS base-calling and consensus-calling steps as containerized workloads that can be scheduled across a pool of compute instances, so a single host failure does not stop a run and capacity can be raised for a batch without a procurement cycle.

    • Hybrid data path from sequencer to cloud

      From device signal to Google Cloud

      Define a secure, low-latency data path from the Oxford Nanopore devices at Matrix ONE into Google Cloud, with schema validation at the boundary so that incomplete or malformed runs fail loudly rather than producing a downstream result that has to be discarded.

    • Elastic capacity for AI training

      GPU capacity that follows the workload

      Connect the Epinn training workload to elastic GPU capacity on Google Cloud, so that model training and evaluation windows can burst to the resources they need for the time they need them, and the cost is matched to the workload rather than to a permanently provisioned cluster.

    • Pipeline uptime is no longer pinned to any single host or cluster.
    • Capacity scales with sequencing volume rather than with hardware procurement.
    • AI training windows can use the GPU resources they need for as long as they need them.

    We've done this before

    See how we do Enterprise AI
  • Digital CDMO

    IT/OT architecture and device integration for Matrix ONE

    We design the IT/OT architecture for the Matrix ONE laboratory, agree the protocol and segmentation standards with Cyclomics before commissioning, and integrate Oxford Nanopore and adjacent laboratory instruments against that architecture from day one.

    • Reference architecture for the facility

      One documented data path

      Specify how sequencing devices, laboratory instruments, line supervision, the LIMS and the cloud pipelines connect, including the network segmentation model, so that every vendor on the project builds toward the same target rather than against their own defaults.

    • OPC UA device integration

      Vendor-neutral device data

      Integrate Oxford Nanopore GridION and MinION devices and adjacent laboratory instruments via OPC UA (Open Platform Communications Unified Architecture) where the device supports it, so device data leaves the equipment in a documented, vendor-neutral form rather than staying inside a closed controller.

    • Segmentation and cybersecurity baseline

      OT zones and conduits from day one

      Define the segmentation zones, conduits and remote-access rules to IEC 62443 (the international standard for industrial automation and control system cybersecurity) before commissioning, so that the cybersecurity posture is set with the network rather than re-argued later against a live production line.

    • Device data is reachable from the rest of the stack from the day Matrix ONE goes live.
    • The reference architecture is reusable as the next site is commissioned.
    • Cybersecurity posture is set with the network, not added under operational pressure later.

    We've done this before

    See how we do Digital CDMO
  • Enterprise AI

    Ontology-driven multi-omics data platform

    We design and implement an ontology-based data platform that defines sample, assay, run, methylation call and structural variant as explicit entities, so that sequence, methylation and structural streams from CyclomicsSeq and from multi-site clinical evaluations load against one model.

    • Shared multi-omics ontology

      One agreed set of terms

      Define sample, assay, run, methylation call and structural variant as explicit entities with agreed relationships, so that a query written once returns comparable answers across all three streams and across UMC Utrecht, Erasmus MC and future study sites.

    • Pipelines from sequencer and clinical sites

      Loading every stream

      Build ingestion for Oxford Nanopore device output, for the methylation-calling pipeline and for the structural-variant pipeline, with schema validation at the boundary so that a malformed record fails loudly instead of silently corrupting the training set.

    • Retrieval layer for the Epinn classifier

      Evidence the model can read

      Expose the unified model through a retrieval layer that the Epinn classifier can query across all three streams, so the model weighs sequence, methylation and structural evidence together rather than reconciling three tables per sample.

    • Three data streams become one queryable evidence base, both for the classifier and for clinical review.
    • Multi-site clinical evaluations align against the same model rather than against per-site mappings.
    • New assays and future study sites attach to the model rather than triggering another migration.

    We've done this before

    See how we do Enterprise AI

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what Cyclomics's own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
Laboratory Automation 35 → 80
CyclomicsSeq library preparation is a manual bench skill held by skilled scientists, which sets the throughput ceiling on commercial kit production. Standing the workflow up as a programmed workstation procedure widens that ceiling while leaving the chemistry untouched.
Data Integration 40 → 85
Sequence, methylation and structural streams from each CyclomicsSeq run are handled separately today, and multi-site clinical evaluations (UMC Utrecht, Erasmus MC) multiply the alignment work. Carrying them into one semantic layer lets the Epinn classifier weigh all three streams together.
Regulatory Compliance 45 → 90
The IVDR transition is publicly named as a use of proceeds for the January 2026 round, and Epinn sits under SaMD regulation. Producing IVDR and SaMD evidence from the same pipeline records that produce the science is the gap that has to close before CE-IVD marking.
Cloud Infrastructure 50 → 85
Cyclomics's acceptance into the Google for Startups Cloud Program was driven by computational needs outgrowing previous infrastructure. A high-availability container architecture on Google Cloud lets pipeline uptime and capacity scale with sequencing volume rather than with hardware procurement.
IT/OT Convergence 30 → 80
Matrix ONE is a greenfield facility. Agreeing protocol standards, segmentation zones and equipment data requirements before Oxford Nanopore and adjacent instruments are commissioned puts interoperability into the design rather than into a later integration project.
Cybersecurity 40 → 85
Acquisition by NASDAQ-listed Lunai Bioworks brings the cybersecurity expectations of a public-market parent onto a research-stage organization. Zero Trust segmentation, identity-based access to genomic data and IEC 62443 zoning for the OT layer are the changes that close the gap.

Check this yourself

Our Service Portal has free self-assessments and market comparisons. These are the ones that line up with what we've read above — no sales call required.

Think we've read this right?

Talk to us

Related reading

This is an independent analysis prepared by A4BEE from publicly available information as of February 2026. It reflects A4BEE's own interpretation and opinion, is not affiliated with, endorsed by, or verified with Cyclomics, and may be incomplete or inaccurate. All company names and trademarks are the property of their respective owners. To request a correction or removal, contact [email protected].