Data & cell states
Start with single-cell atlases and expression data, then explore cell-state representations and foundation models.
03 / CELL STATES & PERTURBATIONS
Virtual cells
Connect data, models, and methods for representing cell states, predicting perturbation responses, and evaluating models.
Perturbations & responses
Connect genetic and drug perturbation data with analysis tools and response-prediction methods.
Evaluation & reproducibility
Examine benchmarks, data splits, and generalization for a specific task to understand which questions a model can address.
Start with these resources
Filter this topic →CZ CELLxGENE Census ↗
A public single-cell RNA data resource. Query cells and genes, then bring selected data into Python or R.
Source · CZ CELLxGENE projectView resource details →Virtual Cells Platform ↗
Explore biological datasets, AI models and task benchmarks through a shared platform for model use and evaluation.
Source · Biohub · Virtual Cells PlatformView resource details →scGPT ↗
Use pretrained models for single-cell analysis, including cell representations, annotation and perturbation prediction.
Source · Wang Lab · scGPT projectView resource details →Pertpy ↗
Analyze single-cell perturbation experiments, organize data and metadata, and compare cellular responses to perturbations.
Source · scverse · Pertpy projectView resource details →Related reading
AI virtual cells: How close are we to growing a cell in a computer?
A review of virtual-cell research, from cell foundation models and mechanistic whole-cell simulations to morphology, metabolism, spatial interactions, tissues and embryos. It surveys public data, code and benchmarks, distinguishing validated capabilities from research goals, with references and practical starting points.