{"entity":{"id":"drug-discovery-roadmap","kind":"roadmap","name":"Drug discovery roadmap: screening in mice → maps of dependency → designing in silico","aka":[],"tldr":"Finding the next cancer drug used to mean testing compounds on mice and cell lines and hoping. It now means mapping which genes each cancer cannot live without, growing a patient's tumour in a dish, and designing molecules on a computer; the job is making those tools predict what happens in people.","summary":"Nine in ten cancer drugs that work in mice fail in humans, and most of the history of drug discovery is the attempt to close that gap. Natural-product screening found vincristine and paclitaxel; target-based discovery, structural biology and high-throughput screening produced the kinase inhibitors. What they could not do was predict which patients a drug would help or which combination would hold.\n\nThe present toolkit attacks that directly. Genome-wide CRISPR screens (DepMap) map the dependencies of a thousand cancer cell lines and expose synthetic-lethal targets; patient-derived organoids and xenografts keep a tumour's biology closer to the patient's; functional testing of drugs on a patient's own cells is being run alongside trials. Structure prediction (AlphaFold 3, Boltz, Chai) and generative design have put the first AI-designed molecules into oncology trials, and perturbation datasets of a hundred million cells are training models that try to predict a drug's effect before the experiment.\n\nThe pace is set by the predictive validity of models, by the half of landmark findings that do not reproduce, by the valley between an academic discovery and a funded programme, and by the secrecy that keeps compound libraries and negative results locked up.","asOf":"2026-09-10","links":[{"label":"DepMap portal","url":"https://depmap.org/portal/"},{"label":"AlphaFold 3 (Nature 2024)","url":"https://doi.org/10.1038/s41586-024-07487-w"},{"label":"Tahoe-100M single-cell perturbation atlas","url":"https://www.tahoebio.ai/"}],"tags":[],"related":["depmap","cancer-models","tcga-gdc","genie","cptac","tahoe-100m","arc-virtual-cell-atlas","virtual-cell","ai-oncology-roadmap","adc-generations","idea-tr2-organoid-coclinical-arms","idea-tr2-reference-model-panels","idea-bio1-in-silico-trials-dose","idea-bio1-multi-organ-chip-tox","idea-tr2-phd-replication-year","idea-tr2-preclinical-negative-registry","idea-bio2-shared-compound-access-pool","idea-bio1-undruggable-market-commitment"],"cancers":[],"sections":["drug-discovery"],"technologies":["crispr-screens","organoids","pdx-models","functional-drug-testing","organoid-guided-therapy-scale","pdac-organoid-pharmacotyping","bh3-profiling","high-throughput-screening-libraries","structural-biology-infrastructure","ai-drug-design","de-novo-protein-design","alphafold3","boltz","chai-1","rfdiffusion","esm3","chemistry42","phenom-2","protac-degrader","molecular-glue-platforms","peptide-drug-conjugate","antisense-sirna","degrader-antibody-conjugate","synthetic-lethality-approaches","proteomics","geneformer","scgpt","state-arc","digital-twins-trials"],"targets":[],"drugs":["vincristine","paclitaxel","dactinomycin","imatinib","sotorasib"],"companies":["isomorphic-labs","insilico-medicine","recursion","exscientia","xaira-therapeutics","generate-biomedicines","chai-discovery","hub-organoids","champions-oncology","sengine","curesponse","araris","bicycle-therapeutics","frontier-medicines","vevo-therapeutics","google-deepmind"],"institutions":["nci","arc-institute"],"pathways":[],"terms":[],"trials":[],"people":[],"bottlenecks":["b-preclinical-models","b-reproducibility","b-translational-valley","b-undruggable-targets","b-ip-collaboration","b-negative-results"],"keyPapers":["paper-abramson-nature"],"journals":[],"dependsOn":[],"notes":[],"steps":[{"era":"1950s-1990s","title":"Screening nature and hoping","description":"The NCI screened tens of thousands of compounds in mouse leukaemias and, from 1990, in a panel of sixty human cell lines. It found the periwinkle alkaloid vincristine, the yew-bark taxane paclitaxel and the antibiotic dactinomycin, and it established the pipeline everyone still uses: cells, then mice, then people. What it could not do was say which people.","refs":["nci","vincristine","paclitaxel","dactinomycin","pdx-models"],"status":"historic"},{"era":"1990s-2015","title":"Targets, structures and high-throughput screens","description":"Cloning the oncogenes gave discovery a target; crystal structures gave it a shape to fit; robotic screening of millions of compounds and later DNA-encoded libraries gave it throughput. Imatinib was the proof. The cost was a generation of drugs that hit their target and did nothing for patients, because the cell-line and xenograft models that selected them did not represent human tumours. Nine in ten oncology drugs entering trials still fail.","refs":["imatinib","high-throughput-screening-libraries","structural-biology-infrastructure","b-preclinical-models","pdx-models"],"status":"historic"},{"era":"2015-2024","title":"Maps of dependency and models closer to the patient","description":"Genome-wide CRISPR knockout screens across a thousand cell lines (DepMap) list which genes each cancer cannot live without, and expose synthetic-lethal pairs such as PRMT5 in MTAP-deleted tumours and WRN in mismatch-repair-deficient ones. Patient-derived organoids keep a tumour's architecture and drug response in a dish; xenograft banks keep it in a mouse; TCGA, GENIE and CPTAC supply the genomes and proteomes to interpret them. The first drugs found this way are now in trials.","refs":["crispr-screens","depmap","synthetic-lethality-approaches","organoids","hub-organoids","pdx-models","champions-oncology","cancer-models","tcga-gdc","genie","cptac","proteomics"],"status":"current"},{"era":"2020-2026","title":"Structure prediction and generative design","description":"AlphaFold turned protein structure into a lookup, and AlphaFold 3 (2024), Boltz and Chai extended it to drug and antibody complexes; RFdiffusion and ESM3 design proteins from scratch. Insilico's generative chemistry produced the first AI-discovered drug to reach phase 2; Isomorphic's first oncology candidate entered trials; Recursion merged with Exscientia to pair image-based biology with design; Xaira launched with over a billion dollars to build discovery around these models. The honest scorecard: faster hit-to-candidate, no approved cancer drug yet.","refs":["alphafold3","boltz","chai-1","chai-discovery","rfdiffusion","esm3","ai-drug-design","chemistry42","insilico-medicine","isomorphic-labs","google-deepmind","recursion","exscientia","phenom-2","xaira-therapeutics","generate-biomedicines"],"status":"current"},{"era":"2015-2026","title":"New modalities as platforms","description":"Discovery is no longer only about small molecules. Degrader and molecular glue platforms remove proteins that cannot be inhibited; ADC linker chemistry (Araris) and bicyclic peptide conjugates (Bicycle) turn a payload into a targeted drug; oligonucleotides silence genes; chemoproteomics finds covalent handles on KRAS. Each platform generates candidates faster than trials can test them, which moves the constraint downstream.","refs":["protac-degrader","molecular-glue-platforms","degrader-antibody-conjugate","araris","bicycle-therapeutics","peptide-drug-conjugate","antisense-sirna","frontier-medicines","sotorasib","adc-generations"],"status":"current"},{"era":"2026-2030","title":"Functional precision medicine","description":"Instead of inferring drug response from genotype, test the drug on the patient's own cells: organoid pharmacotyping in pancreatic cancer, the PARIS organoid screen, tumour fragments kept alive with their vessels, BH3 profiling in leukaemia. The proposal that would make it a field is to grow each trial patient's tumour as organoids and let the results decide which platform arm opens next, with shared reference organoid and xenograft panels so every laboratory tests against the same models.","refs":["functional-drug-testing","organoid-guided-therapy-scale","pdac-organoid-pharmacotyping","sengine","curesponse","bh3-profiling","idea-tr2-organoid-coclinical-arms","idea-tr2-reference-model-panels"],"status":"emerging"},{"era":"2026-2030","title":"Perturbation data at the scale the problem needs","description":"Tahoe-100M measured a hundred million single cells across 1,100 drugs and fifty cancer lines; Arc's Virtual Cell Atlas and State model, Geneformer and scGPT are the attempts to learn from that scale how a cell will respond to a perturbation it has never seen. Rigorous benchmarks show current models barely beat simple baselines on unseen contexts, which is the right kind of bad news: the problem is now measurable. The virtual cell roadmap follows this in detail.","refs":["tahoe-100m","vevo-therapeutics","arc-virtual-cell-atlas","arc-institute","state-arc","geneformer","scgpt","virtual-cell"],"status":"emerging"},{"era":"2030+","title":"In silico first","description":"The long-range bet is that a candidate is designed, its dose chosen and its toxicity screened in silico and on linked human organ chips before the first mouse, and that digital twins reduce the size of the trials that follow. That requires models that generalise, which requires data that reproduce. A funded replication in every cancer biology PhD and a registry for preclinical experiments that did not work are the unglamorous prerequisites.","refs":["idea-bio1-in-silico-trials-dose","idea-bio1-multi-organ-chip-tox","digital-twins-trials","de-novo-protein-design","idea-tr2-phd-replication-year","idea-tr2-preclinical-negative-registry","ai-oncology-roadmap"],"status":"speculative"},{"era":"What sets the pace","title":"Models, reproducibility and the valley of death","description":"Preclinical models still do not predict people, fewer than half of landmark findings reproduce, and most academic discoveries die before anyone tests them in humans because no one funds the step between. Companies hold compound libraries and negative results that would save others years. A shared compound pool for rare cancer researchers and a guaranteed purchase prize for the first drug against a named hard target are two proposals that attack the incentive problem directly.","refs":["b-preclinical-models","b-reproducibility","b-translational-valley","b-ip-collaboration","b-negative-results","b-undruggable-targets","idea-bio2-shared-compound-access-pool","idea-bio1-undruggable-market-commitment"],"status":"current"}],"watch":[]},"route":"/roadmaps/drug-discovery-roadmap/","neighbours":{"collection":[{"id":"genie","kind":"collection","name":"AACR Project GENIE","route":"/collections/genie/"},{"id":"arc-virtual-cell-atlas","kind":"collection","name":"Arc Virtual Cell Atlas","route":"/collections/arc-virtual-cell-atlas/"},{"id":"cancer-models","kind":"collection","name":"Cancer Models (PDCM Finder) & HCMI","route":"/collections/cancer-models/"},{"id":"cptac","kind":"collection","name":"CPTAC (Clinical Proteomic Tumor Analysis Consortium)","route":"/collections/cptac/"},{"id":"depmap","kind":"collection","name":"DepMap (Cancer Dependency Map)","route":"/collections/depmap/"},{"id":"tahoe-100m","kind":"collection","name":"Tahoe-100M","route":"/collections/tahoe-100m/"},{"id":"tcga-gdc","kind":"collection","name":"TCGA / NCI Genomic Data Commons","route":"/collections/tcga-gdc/"}],"roadmap":[{"id":"adc-generations","kind":"roadmap","name":"ADC roadmap: from Mylotarg to bispecific and dual-payload ADCs","route":"/roadmaps/adc-generations/"},{"id":"ai-oncology-roadmap","kind":"roadmap","name":"AI in oncology roadmap: pattern readers → foundation models → agents in the workflow","route":"/roadmaps/ai-oncology-roadmap/"},{"id":"virtual-cell","kind":"roadmap","name":"Virtual cell roadmap: from bulk omics to a predictive model of a cancer cell","route":"/roadmaps/virtual-cell/"}],"idea":[{"id":"idea-bio1-undruggable-market-commitment","kind":"idea","name":"A guaranteed purchase prize for the first drug against a named hard target","route":"/ideas/idea-bio1-undruggable-market-commitment/"},{"id":"idea-tr2-preclinical-negative-registry","kind":"idea","name":"A registry for preclinical experiments that did not work","route":"/ideas/idea-tr2-preclinical-negative-registry/"},{"id":"idea-bio2-shared-compound-access-pool","kind":"idea","name":"A shared compound library that rare cancer researchers can actually use","route":"/ideas/idea-bio2-shared-compound-access-pool/"},{"id":"idea-tr2-phd-replication-year","kind":"idea","name":"Every cancer biology PhD begins with a funded replication of a published finding","route":"/ideas/idea-tr2-phd-replication-year/"},{"id":"idea-tr2-organoid-coclinical-arms","kind":"idea","name":"Grow each trial patient's tumour as organoids to decide which platform arm opens next","route":"/ideas/idea-tr2-organoid-coclinical-arms/"},{"id":"idea-bio1-in-silico-trials-dose","kind":"idea","name":"In silico trials to choose the dose before the first patient","route":"/ideas/idea-bio1-in-silico-trials-dose/"},{"id":"idea-bio1-multi-organ-chip-tox","kind":"idea","name":"Linked human organ chips to predict side effects before people are dosed","route":"/ideas/idea-bio1-multi-organ-chip-tox/"},{"id":"idea-tr2-reference-model-panels","kind":"idea","name":"Shared reference organoid and PDX panels that every lab can test against","route":"/ideas/idea-tr2-reference-model-panels/"}],"section":[{"id":"drug-discovery","kind":"section","name":"Drug Discovery Platforms","route":"/fronts/drug-discovery/"}],"technology":[{"id":"ai-drug-design","kind":"technology","name":"AI-driven drug & target discovery","route":"/technologies/ai-drug-design/"},{"id":"alphafold3","kind":"technology","name":"AlphaFold 3","route":"/technologies/alphafold3/"},{"id":"bh3-profiling","kind":"technology","name":"BH3 profiling (functional apoptosis testing)","route":"/technologies/bh3-profiling/"},{"id":"boltz","kind":"technology","name":"Boltz-1 / Boltz-2 (MIT, open)","route":"/technologies/boltz/"},{"id":"chai-1","kind":"technology","name":"Chai-1 / Chai-2","route":"/technologies/chai-1/"},{"id":"chemistry42","kind":"technology","name":"Chemistry42 and Pharma.AI (Insilico)","route":"/technologies/chemistry42/"},{"id":"crispr-screens","kind":"technology","name":"CRISPR functional genomics","route":"/technologies/crispr-screens/"},{"id":"de-novo-protein-design","kind":"technology","name":"De novo designed protein binders","route":"/technologies/de-novo-protein-design/"},{"id":"degrader-antibody-conjugate","kind":"technology","name":"Degrader-antibody conjugate (DAC)","route":"/technologies/degrader-antibody-conjugate/"},{"id":"digital-twins-trials","kind":"technology","name":"Digital twins and virtual control arms","route":"/technologies/digital-twins-trials/"},{"id":"esm3","kind":"technology","name":"ESM3 (EvolutionaryScale)","route":"/technologies/esm3/"},{"id":"functional-drug-testing","kind":"technology","name":"Functional (ex vivo) drug testing","route":"/technologies/functional-drug-testing/"},{"id":"geneformer","kind":"technology","name":"Geneformer","route":"/technologies/geneformer/"},{"id":"high-throughput-screening-libraries","kind":"technology","name":"High-throughput screening and DNA-encoded libraries","route":"/technologies/high-throughput-screening-libraries/"},{"id":"molecular-glue-platforms","kind":"technology","name":"Molecular glue discovery platforms","route":"/technologies/molecular-glue-platforms/"},{"id":"antisense-sirna","kind":"technology","name":"Oligonucleotide therapeutics","route":"/technologies/antisense-sirna/"},{"id":"organoid-guided-therapy-scale","kind":"technology","name":"Organoid-guided therapy at scale","route":"/technologies/organoid-guided-therapy-scale/"},{"id":"organoids","kind":"technology","name":"Patient-derived organoids","route":"/technologies/organoids/"},{"id":"pdx-models","kind":"technology","name":"Patient-derived xenografts","route":"/technologies/pdx-models/"},{"id":"pdac-organoid-pharmacotyping","kind":"technology","name":"PDAC organoid pharmacotyping","route":"/technologies/pdac-organoid-pharmacotyping/"},{"id":"peptide-drug-conjugate","kind":"technology","name":"Peptide-drug & small-molecule-drug conjugates","route":"/technologies/peptide-drug-conjugate/"},{"id":"phenom-2","kind":"technology","name":"Phenom-2 and Recursion OS","route":"/technologies/phenom-2/"},{"id":"protac-degrader","kind":"technology","name":"PROTACs & molecular glues (targeted protein degradation)","route":"/technologies/protac-degrader/"},{"id":"proteomics","kind":"technology","name":"Proteomics & phosphoproteomics","route":"/technologies/proteomics/"},{"id":"rfdiffusion","kind":"technology","name":"RFdiffusion / RFdiffusion2 and ProteinMPNN (Baker Lab)","route":"/technologies/rfdiffusion/"},{"id":"scgpt","kind":"technology","name":"scGPT","route":"/technologies/scgpt/"},{"id":"state-arc","kind":"technology","name":"State (Arc Institute perturbation model)","route":"/technologies/state-arc/"},{"id":"structural-biology-infrastructure","kind":"technology","name":"Structural biology infrastructure (cryo-EM, synchrotrons, AlphaFold)","route":"/technologies/structural-biology-infrastructure/"},{"id":"synthetic-lethality-approaches","kind":"technology","name":"Synthetic lethality approaches","route":"/technologies/synthetic-lethality-approaches/"}],"drug":[{"id":"dactinomycin","kind":"drug","name":"Dactinomycin (actinomycin D)","route":"/drugs/dactinomycin/"},{"id":"imatinib","kind":"drug","name":"Imatinib","route":"/drugs/imatinib/"},{"id":"paclitaxel","kind":"drug","name":"Paclitaxel / nab-paclitaxel","route":"/drugs/paclitaxel/"},{"id":"sotorasib","kind":"drug","name":"Sotorasib","route":"/drugs/sotorasib/"},{"id":"vincristine","kind":"drug","name":"Vincristine","route":"/drugs/vincristine/"}],"company":[{"id":"araris","kind":"company","name":"Araris Biotech (Taiho)","route":"/companies/araris/"},{"id":"bicycle-therapeutics","kind":"company","name":"Bicycle Therapeutics","route":"/companies/bicycle-therapeutics/"},{"id":"chai-discovery","kind":"company","name":"Chai Discovery","route":"/companies/chai-discovery/"},{"id":"champions-oncology","kind":"company","name":"Champions Oncology","route":"/companies/champions-oncology/"},{"id":"curesponse","kind":"company","name":"Curesponse","route":"/companies/curesponse/"},{"id":"exscientia","kind":"company","name":"Exscientia","route":"/companies/exscientia/"},{"id":"frontier-medicines","kind":"company","name":"Frontier Medicines","route":"/companies/frontier-medicines/"},{"id":"generate-biomedicines","kind":"company","name":"Generate:Biomedicines","route":"/companies/generate-biomedicines/"},{"id":"google-deepmind","kind":"company","name":"Google DeepMind (and Google Research)","route":"/companies/google-deepmind/"},{"id":"hub-organoids","kind":"company","name":"HUB Organoids","route":"/companies/hub-organoids/"},{"id":"insilico-medicine","kind":"company","name":"Insilico Medicine","route":"/companies/insilico-medicine/"},{"id":"isomorphic-labs","kind":"company","name":"Isomorphic Labs","route":"/companies/isomorphic-labs/"},{"id":"recursion","kind":"company","name":"Recursion Pharmaceuticals","route":"/companies/recursion/"},{"id":"sengine","kind":"company","name":"SEngine Precision Medicine","route":"/companies/sengine/"},{"id":"vevo-therapeutics","kind":"company","name":"Vevo Therapeutics","route":"/companies/vevo-therapeutics/"},{"id":"xaira-therapeutics","kind":"company","name":"Xaira Therapeutics","route":"/companies/xaira-therapeutics/"}],"institution":[{"id":"arc-institute","kind":"institution","name":"Arc Institute","route":"/institutions/arc-institute/"},{"id":"nci","kind":"institution","name":"National Cancer Institute (NIH)","route":"/institutions/nci/"}],"bottleneck":[{"id":"b-negative-results","kind":"bottleneck","name":"Failures are hidden","route":"/bottlenecks/b-negative-results/"},{"id":"b-preclinical-models","kind":"bottleneck","name":"Lab models that fail to predict what happens in patients","route":"/bottlenecks/b-preclinical-models/"},{"id":"b-reproducibility","kind":"bottleneck","name":"Preclinical results do not reproduce","route":"/bottlenecks/b-reproducibility/"},{"id":"b-ip-collaboration","kind":"bottleneck","name":"Secrecy and intellectual property block collaboration","route":"/bottlenecks/b-ip-collaboration/"},{"id":"b-undruggable-targets","kind":"bottleneck","name":"The undruggable drivers","route":"/bottlenecks/b-undruggable-targets/"},{"id":"b-translational-valley","kind":"bottleneck","name":"The valley of death between lab and product","route":"/bottlenecks/b-translational-valley/"}],"paper":[{"id":"paper-abramson-nature","kind":"paper","name":"Accurate structure prediction of biomolecular interactions with AlphaFold 3","route":"/key-papers/paper-abramson-nature/"}]}}