CellTuned: AI‑driven Regulatory Genomics
for Precision Therapy
Decoding gene regulation to unlock precision therapy.
About
Born from a decade of research
CellTuned is a platform from the Stein Aerts Lab at VIB-KU Leuven — a leading research group at the intersection of regulatory genomics, single-cell biology, and machine learning. For over a decade, the lab has decoded the genomic regulatory code: mapping enhancer logic across cell types and species, pioneering deep learning models for sequence-to-function prediction, and building widely adopted tools including SCENIC+ and CREsted. CellTuned is the vehicle to translate this scientific foundation into applications for drug discovery.
The Challenge
The Discovery Gap: From Phenotype to Mechanism
Traditional Screening
Phenotypic screening delivered major wins, but much of the low-hanging fruit is exhausted. The field needs a new layer of insight.
Current AI Approaches
Detect patterns across large datasets, but often stop at correlation rather than mechanism, disabling them to resolve disease biology at cell state level.
The Result
Limited insights into causal, regulatory disease drivers — leading to weaker hypotheses, low target confidence, and higher late-stage failure risk.
Mechanistic, cell‑state‑resolved regulatory genomics AI modeling can improve discovery decisions and reduce R&D cost and failure
Why Now
The data is finally here.
Mechanistic AI can act on it.
Single-cell data at scale
Disease-relevant single-cell multiomics datasets have reached the scale, cost, and quality needed to train precise, cell-state-resolved models. Aerts lab, Nature Communications 2026 →
Regulatory genomics tools are mature
A decade of building has made mechanistic, interpretable modelling of gene regulation a practical reality, through tools like SCENIC, SCENIC+, CREsted, and deepSCENIC.
Correlation is not enough
Correlative AI models have shown promise but fall short on mechanism. The field is ready for models that explain why — not just what — delivering actionable, causal hypotheses to drive discovery decisions.
The CellTuned Approach
CellTuned brings mechanistic, cell‑state‑specific AI modeling to precision therapy
Cell
Cell state modelling
Right cell states
Identify and model biology in disease-relevant cellular context — resolving heterogeneity that bulk approaches miss.
Tuned
Mechanistic modelling
Mechanistic model
Move beyond association to disease-driving regulatory insight — understanding why a gene is active, not just that it is.
CellTuned
Modular software platform
Actionable targets & interventions
Translate mechanism into targets, enhancers, and precision therapy strategies ready for experimental validation.
Applications
Gene regulatory insight across the drug development pipeline
CellTuned's regulatory genomics engine applies at multiple stages: from first target hypothesis to clinical decision-making.
Early Discovery
Map disease-driving cell states and screen therapeutic targets through in silico perturbation — before touching a cell.
- In silico perturbation screening across cell states
- Disease cell type mapping
- Genetic variant interpretation in regulatory context
Better target decisions
Drug Discovery
Design synthetic enhancers that drive precise, cell-type-specific gene expression for gene therapy and beyond.
- Synthetic enhancer design and optimization
- Cell-type-specific targeting
- Drug repositioning via regulatory network analysis
Smarter therapeutic design
(Pre-)Clinical Trials
Translate mechanistic regulatory insight into causal biomarkers and patient stratification strategies for trials.
- Causal genetic biomarker identification
- Off-target regulatory effect prediction
- Patient stratification by genetic and cell-state profile
Improved trial precision
Process
How It Works
1
Disease profiling at single-cell resolution
Capture gene expression and chromatin accessibility simultaneously at single-cell resolution in disease-relevant tissues to map the full landscape of cell states.
2
Integrated GRN and S2F modeling
Build gene regulatory networks and sequence-to-function models trained on disease-specific data to learn how the regulatory genome controls each cell state.
3
Identify key diseased cell states
Pinpoint disease-driving cell types and the regulatory programs that define them — connecting the genomic code to observable disease biology.
4
Application modules
Deploy mechanistic insights across three application areas:
Programs
Disease Pipeline
Advancing two parallel tracks across multiple disease areas — from data collection through validation.
Track A — StateTarget
| Disease | Data & AI Training | Target Prioritization | In Vitro Validation | In Vivo Validation |
|---|---|---|---|---|
| AMD | ● | ● | ○ | ○ |
| Melanoma | ● | ● | ○ | ○ |
| Alzheimer’s | ● | ○ | ○ | ○ |
| Parkinson’s | ● | ○ | ○ | ○ |
| Liver Fibrosis / MASH | ● | ○ | ○ | ○ |
| Type 1 and Type 2 diabetes | ● | ○ | ○ | ○ |
| Glioblastoma | ● | ○ | ○ | ○ |
Track B — EnhancerDesign
| Disease | Data & AI Training | In Vitro Validation | In Vivo Validation |
|---|---|---|---|
| Melanoma | ● | ● | ○ |
| Alzheimer’s | ● | ● | ○ |
| Parkinson’s | ● | ○ | ● |
| Liver Fibrosis / MASH | ● | ○ | ○ |
| Type 1 and Type 2 diabetes | ● | ○ | ○ |
| Glioblastoma | ● | ○ | ○ |
Stage completed
Stage not yet reached
Early Access
Partner with us
The regulatory genomics layer your drug pipeline is missing.
Open Source
Software
