Loading
What changed between revision 1 and revision 5, compared over the article's markdown source with its frontmatter — so a change to status, evidence or reversibility shows up as the edit it is.
Older · revision 1
Import· initial corpus import, not an edit
Initial import of content/ai-drug-discovery.md — the filesystem corpus, unchanged. Not an edit.
17.2 kB
Revision 1
Revision 5
lines 8–18 → 8–187 unchanged lines not shown
categories: ["foundations", "longevity"]categories: ["foundations", "longevity"]tags: ["machine learning", "drug development", "generative chemistry", "target discovery", "clinical trials", "pharmacology"]tags: ["machine learning", "drug development", "generative chemistry", "target discovery", "clinical trials", "pharmacology"]summary: "The use of machine learning to select drug targets, generate candidate molecules and predict failure, so far demonstrably faster at making compounds than at making medicines."summary: "The use of machine learning to select drug targets, generate candidate molecules and predict failure, so far demonstrably faster at making compounds than at making medicines."updated: "2026-07-27"updated: "2026-10-28"humanEvidence: "Several AI-derived candidates have reached phase 1 and phase 2 trials in people; as of mid-2026 none has completed phase 3 or been approved, and the largest clinical success is a repurposed existing drug."humanEvidence: "AI-derived small molecules have reached phase 1 and 2 trials, with a first phase 3 begun in 2026; sponsors of an mRNA therapy built on algorithm-selected neoantigens report a phase 3 in melanoma met its endpoints, and no drug from either route is approved."access: "Platforms are proprietary to pharmaceutical companies or licensed to them, and some models and datasets are public; no molecule designed this way is available outside a trial, though AI-suggested uses of approved drugs are."access: "Platforms are proprietary to pharmaceutical companies or licensed to them, and some models and datasets are public; no molecule designed this way is available outside a trial, though AI-suggested uses of approved drugs are."reversibility: "context"reversibility: "context"issues: ["Needs a citable source for the absence of any approved AI-discovered drug.", "The phase 1 to approval rate is uncited; Cook 2014 supports the efficacy-failure point only."]issues: ["Needs a citable source for the absence of any approved AI-discovered drug.", "The phase 1 to approval rate is uncited; Cook 2014 supports the efficacy-failure point only.", "The INTerpath-001 paragraph rests on a topline company announcement; detailed data are not yet published.", "The Anthropic case results are company-reported only; no independent verification exists."]------ ```infobox```infoboxlines 22–29 → 22–293 unchanged lines not shown
{ "label": "Type", "value": "Discovery and design method" }, { "label": "Type", "value": "Discovery and design method" }, { "label": "Main tasks", "value": "Target choice, molecule generation, property prediction" }, { "label": "Main tasks", "value": "Target choice, molecule generation, property prediction" }, { "label": "First candidate in trials", "value": "2020" }, { "label": "First candidate in trials", "value": "2020" }, { "label": "Furthest clinical stage", "value": "Phase 2" }, { "label": "Furthest clinical stage", "value": "Phase 3 (first begun 2026)" }, { "label": "Drugs approved from the approach", "value": "None as of 2026" }, { "label": "Drugs approved from the approach", "value": "None as of August 2026" }, { "label": "Main criticism", "value": "Attrition is a biology problem" }, { "label": "Main criticism", "value": "Attrition is a biology problem" }, { "label": "Readiness", "value": "TRL 5" } { "label": "Readiness", "value": "TRL 5" } ] ]lines 36–43 → 36–456 unchanged lines not shown
chemistry claims are the better supported ones: software now proposes synthesizable, potent-lookingchemistry claims are the better supported ones: software now proposes synthesizable, potent-lookingcompounds against a defined target at a speed no medicinal-chemistry team matches by hand, though howcompounds against a defined target at a speed no medicinal-chemistry team matches by hand, though howmuch of that speed the models themselves explain is disputed. The clinical claims are not settled. Asmuch of that speed the models themselves explain is disputed. The clinical claims are not settled. Asof mid-2026 several AI-derived candidates have entered phase 1 and phase 2 trials, and none hasof August 2026 several AI-derived candidates have entered phase 1 and phase 2 trials, the firstcompleted phase 3 or been approved.small molecule among them has begun phase 3, and none has been approved; the one positive phase 3readout claimed for the field comes from a therapy in which the algorithmic step is antigenselection rather than molecule design. ## What the models actually do## What the models actually do lines 53–59 → 55–619 unchanged lines not shown
**Molecule generation** trains models over SMILES strings, molecular graphs or three-dimensional**Molecule generation** trains models over SMILES strings, molecular graphs or three-dimensionalposes, then steers them with reinforcement learning or conditions them on a binding pocket.poses, then steers them with reinforcement learning or conditions them on a binding pocket.Insilico Medicine's 2019 report on the kinase DDR1 was the first widely noticed demonstration,[Insilico Medicine](https://aiwiki.ai/wiki/insilico_medicine)'s 2019 report on the kinase DDR1 was the first widely noticed demonstration,claiming designed inhibitors in weeks rather than months.[^zhavoronkov2019]claiming designed inhibitors in weeks rather than months.[^zhavoronkov2019] **Property prediction** estimates absorption, metabolism, cardiac ion-channel liability and toxicity**Property prediction** estimates absorption, metabolism, cardiac ion-channel liability and toxicitylines 67–76 → 69–937 unchanged lines not shown
**Structure and synthesis.** Protein structure prediction, covered under [[ai-protein-design]],**Structure and synthesis.** Protein structure prediction, covered under [[ai-protein-design]],opened pocket-based design to proteins with no experimental structure, though benchmark studies ofopened pocket-based design to proteins with no experimental structure, though benchmark studies ofdocking have found predicted structures to perform worse than experimental ones. Retrosynthesisdocking have found predicted structures to perform worse than experimental ones. The companybuilt most directly on AlphaFold is [[isomorphic-labs|Isomorphic Labs]], the Alphabet spin-out fromDeepMind that co-developed AlphaFold 3; despite partnerships with three large drug companies, noneof its molecules had entered a human trial as of October 2026. Retrosynthesismodels propose routes, sometimes coupled to automated synthesis platforms so that design, make andmodels propose routes, sometimes coupled to automated synthesis platforms so that design, make andtest run as a closed loop.test run as a closed loop. **Operational and regulatory work.** Several tasks around the chemistry carry no molecule design atall: compiling manufacturing records, drafting regulatory filings, designing and assessing trials,coordinating work across contract laboratories. These are where developers of general-purposelanguage and agent models have entered the field, by a different route from Alphabet's, which spunits drug-design work out into a separate company. Anthropic reported in October 2026 on researchpartnerships with six drug developers — BigHat Biosciences, Bristol Myers Squibb, Divergence Labs,Enveda, Generate Biomedicines and The Column Group — and on programmes of its own, having acquiredMithril Medicine, a one-person company working on two rare forms of muscular dystrophy whose founderhad used the company's Claude models to coordinate experiments across eleven externallaboratories.[^anthropic2026] Anthropic's stated reasoning is that the bottlenecks are distributedthrough the whole development process rather than concentrated in molecule design. ## Development history## Development history ```timeline```timelinelines 82–88 → 99–1075 unchanged lines not shown
{ "year": "2020", "title": "Antibiotic found by screening model", "text": "The Collins laboratory identifies halicin using a graph neural network, the first widely cited antibiotic hit attributed to deep learning." }, { "year": "2020", "title": "Antibiotic found by screening model", "text": "The Collins laboratory identifies halicin using a graph neural network, the first widely cited antibiotic hit attributed to deep learning." }, { "year": "2021", "title": "AlphaFold2 released", "text": "Predicted structures for most of the human proteome are made public, changing what structure-based design can be attempted." }, { "year": "2021", "title": "AlphaFold2 released", "text": "Predicted structures for most of the human proteome are made public, changing what structure-based design can be attempted." }, { "year": "2021–2022", "title": "AI-derived target and molecule reach humans", "text": "Insilico's ISM001-055, a TNIK inhibitor for idiopathic pulmonary fibrosis, enters first-in-human testing after both target and compound were nominated computationally." }, { "year": "2021–2022", "title": "AI-derived target and molecule reach humans", "text": "Insilico's ISM001-055, a TNIK inhibitor for idiopathic pulmonary fibrosis, enters first-in-human testing after both target and compound were nominated computationally." }, { "year": "2024", "title": "Consolidation", "text": "Exscientia and Recursion merge, combining two of the largest AI-first drug discovery companies after neither had produced an approved medicine." } { "year": "2024", "title": "Consolidation", "text": "Exscientia and Recursion merge, combining two of the largest AI-first drug discovery companies after neither had produced an approved medicine." }, { "year": "2026", "title": "First AI-designed small molecule enters phase 3", "text": "Insilico Medicine begins a 320-patient randomized placebo-controlled phase 3 trial of rentosertib in idiopathic pulmonary fibrosis in China, the furthest any generatively designed small molecule has gone." }, { "year": "2026", "title": "Neoantigen therapy meets phase 3 endpoints", "text": "Merck and Moderna report that intismeran autogene, an individualized mRNA therapy encoding up to 34 algorithm-selected neoantigens, met its recurrence-free survival endpoint with pembrolizumab in resected melanoma." }]]`````` lines 93–111 → 112–1584 unchanged lines not shown
own platforms, which makes it the cleanest available test of the full pipeline. It cleared phase 1,own platforms, which makes it the cleanest available test of the full pipeline. It cleared phase 1,and results from a twelve-week randomized placebo-controlled phase 2a trial in China were publishedand results from a twelve-week randomized placebo-controlled phase 2a trial in China were publishedin 2025: one dose arm showed a mean improvement in forced vital capacity where the placebo armin 2025: one dose arm showed a mean improvement in forced vital capacity where the placebo armdeclined.[^xu2025] That trial was not sized to establish clinical benefit, and the compound had notdeclined.[^xu2025] That trial was not sized to establish clinical benefit. In July 2026 the companyentered phase 3 as of mid-2026.announced the start of a 320-patient, 52-week randomized placebo-controlled phase 3 trial in China,the first phase 3 for a small molecule whose target and structure were both machine-nominated; nodata from it exist.[^insilico2026] The rest of the first wave has behaved like ordinary pharmacology. DSP-1181 was announced in 2020 asThe rest of the first wave has behaved like ordinary pharmacology. DSP-1181 was announced in 2020 asthe first AI-designed molecule to enter a clinical trial and did not progress beyond phase 1; a laterthe first AI-designed molecule to enter a clinical trial and did not progress beyond phase 1; a laterExscientia oncology candidate was also dropped after early data. In 2024 Exscientia merged withExscientia oncology candidate was also dropped after early data. In 2024 Exscientia merged withRecursion, whose approach centred on high-throughput cellular imaging rather than generative[Recursion](https://aiwiki.ai/wiki/recursion_pharmaceuticals), whose approach centred on high-throughput cellular imaging rather than generativechemistry. Neither company had an approved drug at the time.chemistry. Neither company had an approved drug at the time. The neighbouring field of computational protein design has produced no approved medicine of its ownThe neighbouring field of computational protein design has produced no approved medicine of its owneither. A designed nanoparticle scaffold is a component of a COVID-19 vaccine approved in Southeither. A designed nanoparticle scaffold is a component of a COVID-19 vaccine approved in SouthKorea, but as of 2026 no de novo designed protein has been approved anywhere as a therapeutic drug.Korea, but as of 2026 no de novo designed protein has been approved anywhere as a therapeutic drug. A phase 3 result of a different kind arrived in August 2026, from outside small-molecule chemistry.Merck and Moderna announced that intismeran autogene, an individualized neoantigen mRNA therapycovered under [[personalized-mrna-cancer-vaccines]], met its primary endpoint of recurrence-freesurvival and a key secondary endpoint of distant metastasis-free survival when added topembrolizumab in a 1,137-patient trial in resected stage IIB–IV melanoma. The claim is topline, froman interim analysis, and detailed data had not been released as of late August 2026.[^merck2026]Each course is built by sequencing the patient's tumour and encoding up to 34 of its mutations,selected by what Moderna describes as a proprietary algorithm developed with Merck.[^moderna2023]That computational step ranks which of a patient's existing mutations to encode rather thangenerating a new chemical entity, so whether the result counts as an AI-derived drug succeeding inphase 3 depends entirely on where the label's boundary is drawn. The newest claims for the approach are not near patients at all. Anthropic's October 2026 reportmixes molecule-design results — roughly 35,000 antibody designs generated for Divergence Labs inabout a month, about 1,010 antibody variants made and tested with BigHat Biosciences over about tenweeks, a shortened genetic component of a CAR T-cell therapy that Generate Biomedicines aims toextend to [[cancer|solid tumours]] — with operational ones, among them a manufacturing-documentreview at Enveda that the company says cut human review time by 89%, and an assessment ofclinical-trial plans at Bristol Myers Squibb. Every figure is reported by the companies themselves,and Anthropic describes the work as early and mostly preclinical: nothing from it has entered ahuman trial, and the earliest first-in-human date the company offers, late 2028, is conditional onits candidates continuing to hold up. Its furthest-advanced internal programme, a bispecificantibody for a form of congenital muscular dystrophy, rests on a pilot study in mice whoseaccompanying technical report states that it was not peer reviewed and that the experiments wereneither randomized nor blinded. > [!caution] What "AI-discovered" means> [!caution] What "AI-discovered" means> No regulator recognizes the category, and companies apply the label to everything from a molecule> No regulator recognizes the category, and companies apply the label to everything from a molecule> drawn by a generative model to a conventional campaign in which a classifier filtered one plate.> drawn by a generative model to a conventional campaign in which a classifier filtered one plate.lines 222–228 → 269–279110 unchanged lines not shown
[^richardson2020]: `paper` Richardson, P. et al. "Baricitinib as potential treatment for 2019-nCoV acute respiratory disease." *The Lancet*, 2020. {Baricitinib was already approved for rheumatoid arthritis, so the system reranked known pharmacology rather than proposing a new molecule.}[^richardson2020]: `paper` Richardson, P. et al. "Baricitinib as potential treatment for 2019-nCoV acute respiratory disease." *The Lancet*, 2020. {Baricitinib was already approved for rheumatoid arthritis, so the system reranked known pharmacology rather than proposing a new molecule.}[^zhavoronkov2019]: `paper` Zhavoronkov, A. et al. "Deep learning enables rapid identification of potent DDR1 kinase inhibitors." *Nature Biotechnology*, 2019. {DDR1 was a well-precedented target and the compounds resembled known inhibitors; chemists disputed how much of the speed the model explained.}[^zhavoronkov2019]: `paper` Zhavoronkov, A. et al. "Deep learning enables rapid identification of potent DDR1 kinase inhibitors." *Nature Biotechnology*, 2019. {DDR1 was a well-precedented target and the compounds resembled known inhibitors; chemists disputed how much of the speed the model explained.}[^stokes2020]: `paper` Stokes, J.M. et al. "A Deep Learning Approach to Antibiotic Discovery." *Cell*, 2020. {Halicin's activity was confirmed in mouse infection models; it has not entered human trials.}[^stokes2020]: `paper` Stokes, J.M. et al. "A Deep Learning Approach to Antibiotic Discovery." *Cell*, 2020. {Halicin's activity was confirmed in mouse infection models; it has not entered human trials.}[^anthropic2026]: `statement` Anthropic. "Early progress on accelerating drug development with Claude." Company blog post, October 27, 2026, with the accompanying technical report by Tranbarger, C. and Mithril Medicine on a bispecific bridging antibody for LAMA2-related congenital muscular dystrophy, October 26, 2026. {Every result is the companies' own; the technical report is marked not peer reviewed and its mouse pilot was neither randomized nor blinded.}[^xu2025]: `paper` Xu, Z. et al. "A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis: a randomized phase 2a trial." *Nature Medicine*, 2025. {Seventy-one patients across three dose arms and placebo, treated for twelve weeks; a trial that size and length cannot establish clinical benefit in fibrosis.}[^xu2025]: `paper` Xu, Z. et al. "A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis: a randomized phase 2a trial." *Nature Medicine*, 2025. {Seventy-one patients across three dose arms and placebo, treated for twelve weeks; a trial that size and length cannot establish clinical benefit in fibrosis.}[^insilico2026]: `statement` Insilico Medicine. "Insilico Initiates Phase III Clinical Trial for Rentosertib, Its AI-Empowered TNIK Inhibitor for Idiopathic Pulmonary Fibrosis." Press release, July 2026. {Announces trial initiation only, not results; a 320-patient, 52-week randomized placebo-controlled study at Chinese centres.}[^merck2026]: `statement` Merck and Moderna. "Merck and Moderna Announce Phase 3 INTerpath-001 Trial of Intismeran Autogene Plus KEYTRUDA Met Endpoints of RFS and DMFS in Patients With Completely Resected Stage IIB–IV Melanoma." Joint press release, August 19, 2026. {A topline interim-analysis announcement; hazard ratios and detailed results were not disclosed with it.}[^moderna2023]: `statement` Moderna. "Exploring One Medicine for One Patient: Individualized Neoantigen Therapies." Company blog post, 2023. {The company's own description of its pipeline; the selection algorithm's architecture and validation are not publicly disclosed.}[^cook2014]: `paper` Cook, D. et al. "Lessons learned from the fate of AstraZeneca's drug pipeline: a five-dimensional framework." *Nature Reviews Drug Discovery*, 2014.[^cook2014]: `paper` Cook, D. et al. "Lessons learned from the fate of AstraZeneca's drug pipeline: a five-dimensional framework." *Nature Reviews Drug Discovery*, 2014.[^nelson2015]: `paper` Nelson, M.R. et al. "The support of human genetic evidence for approved drug indications." *Nature Genetics*, 2015. {The association is between genetic support and eventual approval; it does not show that adding genetics to a programme causes success.}[^nelson2015]: `paper` Nelson, M.R. et al. "The support of human genetic evidence for approved drug indications." *Nature Genetics*, 2015. {The association is between genetic support and eventual approval; it does not show that adding genetics to a programme causes success.}[^chen2019]: `paper` Chen, L. et al. "Hidden bias in the DUD-E dataset leads to misleading performance of deep learning in structure-based virtual screening." *PLoS ONE*, 2019. {The bias sits in the benchmark rather than in any one model, which makes published virtual-screening gains hard to compare.}[^chen2019]: `paper` Chen, L. et al. "Hidden bias in the DUD-E dataset leads to misleading performance of deep learning in structure-based virtual screening." *PLoS ONE*, 2019. {The bias sits in the benchmark rather than in any one model, which makes published virtual-screening gains hard to compare.}removed, struck through added, underlinedLine numbers count the serialised markdown of each revision, frontmatter included.
219 lines are the same in both revisions, and 155 of them are not shown. Neither revision here is necessarily the current article — read AI drug discovery for the text this wiki stands behind today.