{"id":"virtual-cell","name":"Virtual cell roadmap: from bulk omics to a predictive model of a cancer cell","route":"/roadmaps/virtual-cell/","eras":[{"era":"2008-2018","title":"Atlases and bulk omics","description":"TCGA catalogues the genomes of 11,000 tumours; single-cell RNA-seq matures; the Human Cell Atlas begins. Models are statistical, per-dataset, and descriptive.","status":"historic","refs":[{"id":"tcga-gdc","kind":"collection","name":"TCGA / NCI Genomic Data Commons","route":"/collections/tcga-gdc/","tldr":"The reference atlas of cancer genomes that most cancer biology since 2008 is built on."},{"id":"cellxgene-hca","kind":"collection","name":"CZ CELLxGENE / Human Cell Atlas","route":"/collections/cellxgene-hca/","tldr":"CELLxGENE and the Human Cell Atlas hold single-cell data from healthy and diseased tissue, browsable and downloadable."},{"id":"single-cell-spatial","kind":"technology","name":"Single-cell & spatial profiling","route":"/technologies/single-cell-spatial/","status":"emerging","tldr":"Reading the genes of each individual cell, and mapping where each cell sits in the tumour."}],"trials":[],"papers":[]},{"era":"2019-2022","title":"Pooled perturbation screens meet single cells","description":"Perturb-seq and genome-wide CRISPR screens (DepMap) give causal training data; GEARS shows graph models can predict some unseen knockouts.","status":"historic","refs":[{"id":"crispr-screens","kind":"technology","name":"CRISPR functional genomics","route":"/technologies/crispr-screens/","status":"established","tldr":"Knocking out every gene one at a time in cancer cells to find which ones they cannot live without."},{"id":"depmap","kind":"collection","name":"DepMap (Cancer Dependency Map)","route":"/collections/depmap/","tldr":"Which genes each cancer cell line cannot live without. The map of synthetic-lethal targets."},{"id":"gears","kind":"technology","name":"GEARS and perturbation prediction benchmarks","route":"/technologies/gears/","status":"emerging","tldr":"GEARS is a graph model predicting the effect of gene knockouts; the perturbation benchmarks around it showed how hard the problem is."}],"trials":[],"papers":[]},{"era":"2023-2024","title":"First single-cell foundation models","description":"Geneformer, scGPT, UCE, scFoundation and others pretrain on tens of millions of cells. Benchmarks reveal that perturbation prediction often does not beat linear or mean baselines, forcing better evaluation.","status":"current","refs":[{"id":"geneformer","kind":"technology","name":"Geneformer","route":"/technologies/geneformer/","status":"emerging","tldr":"Geneformer is a transformer trained on about 30 million single cells that encodes each cell as a ranked list of its genes, so deleting a gene in silico shows which genes matter in a disease. It was the first single-cell foundation model in general use, though benchmarks find only modest gains over linear baselines on some tasks."},{"id":"scgpt","kind":"technology","name":"scGPT","route":"/technologies/scgpt/","status":"emerging","tldr":"A GPT-style model for single-cell data that predicts cell types, perturbation responses, and gene networks."},{"id":"universal-cell-embedding","kind":"technology","name":"Universal Cell Embedding (UCE)","route":"/technologies/universal-cell-embedding/","status":"emerging","tldr":"Universal Cell Embedding maps any cell from any species into one shared space without retraining."},{"id":"scfoundation","kind":"technology","name":"scFoundation (BioMap)","route":"/technologies/scfoundation/","status":"emerging","tldr":"scFoundation is a 100-million-parameter model trained on 50 million cells, from China's BioMap."},{"id":"gears","kind":"technology","name":"GEARS and perturbation prediction benchmarks","route":"/technologies/gears/","status":"emerging","tldr":"GEARS is a graph model predicting the effect of gene knockouts; the perturbation benchmarks around it showed how hard the problem is."}],"trials":[],"papers":[]},{"era":"2025-2026","title":"Data at scale and context-aware models","description":"Tahoe-100M (100M cells, 1,100 drugs, 50 cancer lines), Arc's Virtual Cell Atlas and Challenge, State trained on 100M+ perturbed cells, C2S-Scale's lab-validated hypothesis, CZI's cross-species models. The problem becomes one of held-out generalisation across cell contexts.","status":"current","refs":[{"id":"tahoe-100m","kind":"collection","name":"Tahoe-100M","route":"/collections/tahoe-100m/","tldr":"Tahoe-100M is the biggest single-cell dataset ever released, built to teach AI how cancer cells respond to drugs."},{"id":"arc-virtual-cell-atlas","kind":"collection","name":"Arc Virtual Cell Atlas","route":"/collections/arc-virtual-cell-atlas/","tldr":"Arc's growing library of cell data, the fuel for virtual cell models."},{"id":"state-arc","kind":"technology","name":"State (Arc Institute perturbation model)","route":"/technologies/state-arc/","status":"emerging","tldr":"Predicts how cells will respond to a drug or gene knockout, trained on over 100 million perturbed cells."},{"id":"c2s-scale","kind":"technology","name":"Cell2Sentence / C2S-Scale (Yale, Google)","route":"/technologies/c2s-scale/","status":"emerging","tldr":"Turns a cell's gene expression into a sentence so a normal language model can reason about it; a 27-billion-parameter version proposed a cancer immunotherapy idea that was confirmed in the lab."},{"id":"transcriptformer","kind":"technology","name":"TranscriptFormer and rBio (CZI virtual cell models)","route":"/technologies/transcriptformer/","status":"emerging","tldr":"CZI's open cross-species cell models and a reasoning model trained on them."},{"id":"nicheformer","kind":"technology","name":"Nicheformer (spatial single-cell)","route":"/technologies/nicheformer/","status":"emerging","tldr":"Nicheformer is a model trained on both dissociated and spatial data so it learns how a cell's neighbourhood shapes it."}],"trials":[],"papers":[]},{"era":"2027-2029","title":"Patient-derived contexts and spatial niches","description":"Models trained on perturbations in patient-derived organoids and spatial data (tumour niches, immune contexts) rather than cell lines alone; coupling with structure models for mechanism; prospective use to rank drug combinations for organoid confirmation.","status":"emerging","refs":[{"id":"organoids","kind":"technology","name":"Patient-derived organoids","route":"/technologies/organoids/","status":"established","tldr":"Patient-derived organoids are miniature 3D versions of a patient's tumour grown in the lab."},{"id":"functional-drug-testing","kind":"technology","name":"Functional (ex vivo) drug testing","route":"/technologies/functional-drug-testing/","status":"emerging","tldr":"Growing a patient's own cancer cells in a dish and testing drugs on them directly, instead of guessing from genetics."},{"id":"nicheformer","kind":"technology","name":"Nicheformer (spatial single-cell)","route":"/technologies/nicheformer/","status":"emerging","tldr":"Nicheformer is a model trained on both dissociated and spatial data so it learns how a cell's neighbourhood shapes it."},{"id":"boltz","kind":"technology","name":"Boltz-1 / Boltz-2 (MIT, open)","route":"/technologies/boltz/","status":"emerging","tldr":"Open-source structure models that match AlphaFold 3, with Boltz-2 also predicting how strongly a drug binds."}],"trials":[],"papers":[]},{"era":"2030+","title":"Speculative: in silico trials and digital twins","description":"A tumour's multi-omic profile seeds a patient-specific virtual cell population; treatment sequences are simulated before the first cycle; models are updated from ctDNA and imaging during care. Requires validation standards that do not yet exist.","status":"speculative","refs":[{"id":"idea-multimodal-foundation-model","kind":"idea","name":"Patient-level multimodal foundation models for treatment selection","route":"/ideas/idea-multimodal-foundation-model/","tldr":"Train one AI on scans, pathology slides, genomics and treatment outcomes pooled across patients, including completed phase 3 trials, so it can predict which treatment will work for a new patient. Pathology and radiology models already exist separately; combining them with genomic and trial outcome data is the untested step."},{"id":"mrd-testing","kind":"technology","name":"MRD / molecular residual disease testing","route":"/technologies/mrd-testing/","status":"established","tldr":"An ultra-sensitive blood test after surgery that detects leftover cancer months before a scan would."}],"trials":[],"papers":[]}],"watch":[]}