Nygen Analytics Announces the Release of CyteType

A Multi-LLM Platform for High-Fidelity Cell Annotation and Characterization

By
Sukhitha Basnayake
30 May 2025
Share this post

Nygen Analytics Introduces CyteType® A New Standard in Cell Annotation for Drug Discovery

Malmö, Sweden | 30.05.2025 - Nygen Analytics has launched CyteType®, a Python-based multi-agentic AI platform designed to elevate how scientists interpret single-cell data in disease-focused research. While most cell annotation tools assign simple labels to clusters, CyteType® acts more like an “AI biologist” offering contextual biological interpretation, diagnostic reasoning, and peer-review-style analysis. It is purpose-built to accelerate cell characterization workflows in academic and industrial pipelines, including immuno-oncology, inflammation, and hematological malignancies.

At the core of CyteType’s innovation is its multi-agent LLM architecture, combining models such as Anthropic’s Claude Sonnet 4, xAI’s Grok 3, and OpenAI’s GPT-4 (04). These reasoning-capable models work in concert to generate not just annotations—but interpretations. Each cluster is evaluated through a dynamic set of tools and ontologies, with predictions grounded in the latest literature and standardized against authoritative biological references.

“What sets CyteType apart is that it doesn't just answer ‘what is this cell?’ it answers ‘why is it behaving this way in this context?’” said Göran Karlsson, Associate Professor of Molecular Hematology and Co-founder of Nygen Analytics. “This is critical in disease settings where biological signals are complex, noisy, and highly contextual.”

A Shift From Annotation to Interpretation

Standard tools such as Seurat, SingleR, and CellTypist provide a foundational layer of automated annotation labeling clusters based on gene expression profiles. CyteType® builds on this by integrating biological reasoning, disease-awareness, and narrative synthesis. In practical terms:

  • A standard tool might output:
    Cluster 1: Plasma cell
    Top markers: IGKC, JCHAIN, MZB1
  • CyteType provides instead:
    “Activated, antibody-secreting plasma cells. Expression of IGHA1 suggests mucosal-associated subset. Alternative interpretations (e.g., plasmablastic lymphoma) considered and ruled out. Capture efficiency considerations noted.”

The result is a comprehensive, peer-review-style report for every dataset where methodological caveats, unexpected expression patterns, and disease-specific insights are clearly laid out for the scientist to evaluate or build upon.

Designed for Biopharma Workflows

CyteType’s modular architecture integrates seamlessly into Python-based single-cell pipelines such as Scanpy and AnnData workflows. Its outputs are structured, reproducible, and exportable suitable for downstream analyses, manuscript preparation, and collaborative research across teams. It supports both cloud-based and local deployments, with minimal setup friction.

Key design principles for biopharma teams:

  • Explainability: Every annotation includes justification and reference-standard validation.
  • Scalability: Efficient on large datasets; adaptable to high-throughput pipelines.
  • Modularity: Supports tool integration and model replacement; version-controlled.
  • Security-ready: Can operate in air-gapped or firewall-restricted environments.

A Community-Driven Future

CyteType is open-source and available now at github.com/NygenAnalytics/CyteType. Nygen invites feedback and collaboration from the scientific community to further expand tissue coverage, improve rare cell identification, and enhance interpretability features across disease areas.

About Nygen Analytics
Nygen Analytics develops AI-native platforms for interpreting single-cell and spatial omics data. Combining domain-specialized LLMs, efficient data structures, and a deep understanding of translational biology, Nygen supports biopharma and academic researchers in uncovering actionable insights from high-dimensional data.

Press Contact
sukhitha.basnayake@nygen.io
www.nygen.io

Sukhitha Basnayake
Scientific Commmunitcations Manager