{"slug": "alphafold", "name": "AlphaFold", "category": "ai-models", "category_name": "AI models", "description": "", "recommenders": 13, "commenters": 21, "warners": 4, "score": 8.898, "rising": -0.27, "first": "2025-11-30", "latest": "2026-09-06", "why": "AlphaFold is great...to come up with a starting model for a chimeric fusion or something.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 13, "neutral": 3, "compare": 1, "warn": 4}, "spark": [1, 2, 0, 3, 0, 0, 0, 3, 1, 2, 1, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "AlphaFold is great...to come up with a starting model for a chimeric fusion or something.", "kind": "", "icon": true, "domain": "ac.uk", "github_repo": null, "official_url": "https://deepmind.google/science/alphafold/", "pricing": "", "aliases": ["AlphaFold", "AlphaFold 3", "AlphaFold2", "AlphaFold 2"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 793444992, "item_id": 49590312, "url": "https://news.ycombinator.com/item?id=49590312", "by": "Fraterkes", "date": "2026-09-06", "version": null, "stance": "recommend", "quote": "AlphaFold being an exception", "excerpt": "What I'd like these people to (publicly) grapple with is the following: The results of the past few years of ai development have been disruptive largely in the area of white-collar work. Comparatively the results in ie ai-enabled medical advancements have been modest (AlphaFold being an exception); I think it's telling that the main achievement touted here is providing people with cheap medical counseling. So if we p", "context": "", "use_case": "", "switched_from": "", "story": "An Alien Mind"}, {"id": 789033312, "item_id": 49314582, "url": "https://news.ycombinator.com/item?id=49314582", "by": "colingauvin", "date": "2026-08-15", "version": null, "stance": "recommend", "quote": "AlphaFold is great...to come up with a starting model for a chimeric fusion or something.", "excerpt": "I'm a structural biologist at a mid-sized biotech. I use AI tools daily. They make accomplishing the same things I was able to accomplish before quite a lot faster and easier. They don't help me magically accomplish new things that I couldn't previously. For example, it helps me install academic software, debug things. It helps me take a large dataset and write scripts to ask questions. It helps me go through experim", "context": "structural biologist at a mid-sized biotech", "use_case": "starting models for protein fusions", "switched_from": "", "story": "So How Is AI Drug Discovery Doing, Really?"}, {"id": 787061712, "item_id": 49191357, "url": "https://news.ycombinator.com/item?id=49191357", "by": "est31", "date": "2026-08-05", "version": null, "stance": "recommend", "quote": "protein folding has been figured out by AI", "excerpt": "For example, protein folding has been figured out by AI. This was a very big moment for science and yielded a nobel prize. Alphafold 1 happened 4 years before the first version of ChatGPT and the LLM craze we see today. But yeah, most problems in physics, chemistry or biology require labs on top of actual hard thinking. You need to be able to design experiments in a certain way. Once you have the funding, the right t", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "Why the Legendary Erd\u0151s Problems Are Falling to AI"}, {"id": 785960768, "item_id": 49122548, "url": "https://news.ycombinator.com/item?id=49122548", "by": "ben_w", "date": "2026-07-31", "version": null, "stance": "recommend", "quote": "Deep learning: https://en.wikipedia.org/wiki/AlphaFold", "excerpt": "> Can you give a few specific examples? Early AI, the \"Symbolic AI/Expert systems\" era: https://en.wikipedia.org/wiki/Dendral Evolutionary computation era: https://en.wikipedia.org/wiki/Evolved_antenna Deep learning: https://en.wikipedia.org/wiki/AlphaFold Generative AI: https://openai.com/index/model-disproves-discrete-geometry-conjecture/ but also I've lost track of how many Erd\u0151s problems have been solved in the l", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "Is AI Capable of Novelty?"}, {"id": 777273664, "item_id": 48579604, "url": "https://news.ycombinator.com/item?id=48579604", "by": "astrange", "date": "2026-06-17", "version": null, "stance": "recommend", "quote": "I like how you're trying to argue with the Nobel committee here.", "excerpt": "https://www.the-scientist.com/chatgpt-and-alphafold-help-design-personalized-vaccine-for-dog-with-cancer-74227 https://arxiv.org/pdf/2503.15204 > They already fold proteins with simulations. I like how you're trying to argue with the Nobel committee here.", "context": "", "use_case": "", "switched_from": "", "story": "Leaked financial docs show OpenAI is losing billions of dollars a year"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 789569760, "item_id": 49348110, "url": "https://news.ycombinator.com/item?id=49348110", "by": "fizzbarnull", "date": "2026-08-18", "version": null, "stance": "warn", "quote": "these folding models fail when we don't have crystal structures", "excerpt": "> Simulating biology is what we were doing with protein folding before Alphafold This is not correct, Alphafold predicts crystal structures of proteins. It is a known issue in the field that these folding models do not generate ensembles of protein conformations like the ensembles generated by DE Shaw Research (who built a super computer to simulate proteins). And these folding models fail when we don't have crystal ", "context": "", "use_case": "", "switched_from": "", "story": "Amodei Re. Regulation"}, {"id": 774205728, "item_id": 48387858, "url": "https://news.ycombinator.com/item?id=48387858", "by": "photochemsyn", "date": "2026-06-03", "version": null, "stance": "warn", "quote": "This does reveal the weakness of AlphaFold approaches", "excerpt": "This does reveal the weakness of AlphaFold approaches for answering questions like \u201cwhat is possible in the protein folding space if you use the 20 canonical amino acids\u201d since the data used to train AlphaFold is limited to existing experimentally determined protein structures. We don\u2019t even know if this is like body plans (four legs for mammals, why not six?) i.e. is this about physical limitations of the folding sp", "context": "", "use_case": "", "switched_from": "", "story": "The Unreasonable Redundancy of Nature's Protein Folds"}, {"id": 772533298, "item_id": 48283331, "url": "https://news.ycombinator.com/item?id=48283331", "by": "Subtextofficial", "date": "2026-05-26", "version": null, "stance": "warn", "quote": "hours on AlphaFold queue", "excerpt": "Built this in a week on owned hardware (4x GPU, 32 shards). Running on a single machine in Portland with $126/mo overhead. Live endpoints: - Emotion detection: 16ms P50, 10x faster than Azure at same price - CRISPR guide RNA design: only self-serve API, Benchling charges $20k+/year - Protein folding: ESMFold, sub-10s vs hours on AlphaFold queue - Local LLM: 97% cheaper than OpenAI, data never leaves the server - Town", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 766741425, "item_id": 47921339, "url": "https://news.ycombinator.com/item?id=47921339", "by": "squid-protocol", "date": "2026-04-27", "version": null, "stance": "warn", "quote": "It revealed a very linear data pipeline stuffed with steps of O(N)^6 algorithms.", "excerpt": "I've been working on a custom AST-free, LLM-free code base analysis engine and used it to generate the following data to dissect Alphafold. It revealed a very linear data pipeline stuffed with steps of O(N)^6 algorithms. repo - https://github.com/squid-protocol/gitgalaxy", "context": "", "use_case": "", "switched_from": "", "story": "Teardown of AlphaFold codebase with a AST-free LLM-free static analysis engine"}]}], "who": [{"by": "colingauvin", "context": "structural biologist at a mid-sized biotech", "url": "https://news.ycombinator.com/item?id=49314582", "date": "2026-08-15"}], "tags": [], "use_cases": ["protein structure prediction", "starting models for protein fusions", "dog cancer vaccine design"], "compared": [{"slug": "chatgpt", "name": "ChatGPT", "co": 3, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=46270972", "title": "Why proteins fold and how GPUs help us fold", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=47120201", "title": "SETI@home: Data Acquisition and Front-End Processing", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49588080", "title": "An Alien Mind", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49325789", "title": "Amodei Re. Regulation", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49313367", "title": "So How Is AI Drug Discovery Doing, Really?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49261532", "title": "Grok Bot by SpaceXAI", "mentions": 1}], "sentiment_by_month": {"2025-11": {"recommend": 1}, "2025-12": {"recommend": 2}, "2026-01": {}, "2026-02": {"recommend": 3, "compare": 1}, "2026-03": {}, "2026-04": {"warn": 1}, "2026-05": {"warn": 1}, "2026-06": {"recommend": 3, "warn": 1}, "2026-07": {"neutral": 2, "recommend": 1}, "2026-08": {"recommend": 2, "neutral": 1, "warn": 1}, "2026-09": {"recommend": 1}, "2026-10": {}}, "versions": [], "version_unnamed": 17, "family_ref": null, "runs_on": null, "lines": [], "quotes": [{"id": 793444992, "item_id": 49590312, "url": "https://news.ycombinator.com/item?id=49590312", "by": "Fraterkes", "date": "2026-09-06", "version": null, "stance": "recommend", "quote": "AlphaFold being an exception", "excerpt": "What I'd like these people to (publicly) grapple with is the following: The results of the past few years of ai development have been disruptive largely in the area of white-collar work. Comparatively the results in ie ai-enabled medical advancements have been modest (AlphaFold being an exception); I think it's telling that the main achievement touted here is providing people with cheap medical counseling. So if we p", "context": "", "use_case": "", "switched_from": "", "story": "An Alien Mind"}, {"id": 789569760, "item_id": 49348110, "url": "https://news.ycombinator.com/item?id=49348110", "by": "fizzbarnull", "date": "2026-08-18", "version": null, "stance": "warn", "quote": "these folding models fail when we don't have crystal structures", "excerpt": "> Simulating biology is what we were doing with protein folding before Alphafold This is not correct, Alphafold predicts crystal structures of proteins. It is a known issue in the field that these folding models do not generate ensembles of protein conformations like the ensembles generated by DE Shaw Research (who built a super computer to simulate proteins). And these folding models fail when we don't have crystal ", "context": "", "use_case": "", "switched_from": "", "story": "Amodei Re. Regulation"}, {"id": 789033312, "item_id": 49314582, "url": "https://news.ycombinator.com/item?id=49314582", "by": "colingauvin", "date": "2026-08-15", "version": null, "stance": "recommend", "quote": "AlphaFold is great...to come up with a starting model for a chimeric fusion or something.", "excerpt": "I'm a structural biologist at a mid-sized biotech. I use AI tools daily. They make accomplishing the same things I was able to accomplish before quite a lot faster and easier. They don't help me magically accomplish new things that I couldn't previously. For example, it helps me install academic software, debug things. It helps me take a large dataset and write scripts to ask questions. It helps me go through experim", "context": "structural biologist at a mid-sized biotech", "use_case": "starting models for protein fusions", "switched_from": "", "story": "So How Is AI Drug Discovery Doing, Really?"}, {"id": 788926112, "item_id": 49307882, "url": "https://news.ycombinator.com/item?id=49307882", "by": "no-name-here", "date": "2026-08-14", "version": null, "stance": "neutral", "quote": "Does AlphaFold since 2018 (from Google, and now as the Google spin-off Isomorphic Labs under Demis Hassabis) count?", "excerpt": "Does AlphaFold since 2018 (from Google, and now as the Google spin-off Isomorphic Labs under Demis Hassabis) count? https://deepmind.google/science/alphafold/", "context": "", "use_case": "", "switched_from": "", "story": "Grok Bot by SpaceXAI"}, {"id": 787061712, "item_id": 49191357, "url": "https://news.ycombinator.com/item?id=49191357", "by": "est31", "date": "2026-08-05", "version": null, "stance": "recommend", "quote": "protein folding has been figured out by AI", "excerpt": "For example, protein folding has been figured out by AI. This was a very big moment for science and yielded a nobel prize. Alphafold 1 happened 4 years before the first version of ChatGPT and the LLM craze we see today. But yeah, most problems in physics, chemistry or biology require labs on top of actual hard thinking. You need to be able to design experiments in a certain way. Once you have the funding, the right t", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "Why the Legendary Erd\u0151s Problems Are Falling to AI"}, {"id": 785960768, "item_id": 49122548, "url": "https://news.ycombinator.com/item?id=49122548", "by": "ben_w", "date": "2026-07-31", "version": null, "stance": "recommend", "quote": "Deep learning: https://en.wikipedia.org/wiki/AlphaFold", "excerpt": "> Can you give a few specific examples? Early AI, the \"Symbolic AI/Expert systems\" era: https://en.wikipedia.org/wiki/Dendral Evolutionary computation era: https://en.wikipedia.org/wiki/Evolved_antenna Deep learning: https://en.wikipedia.org/wiki/AlphaFold Generative AI: https://openai.com/index/model-disproves-discrete-geometry-conjecture/ but also I've lost track of how many Erd\u0151s problems have been solved in the l", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "Is AI Capable of Novelty?"}, {"id": 785807392, "item_id": 49112962, "url": "https://news.ycombinator.com/item?id=49112962", "by": "tantalor", "date": "2026-07-30", "version": null, "stance": "neutral", "quote": "Google DeepMind disbands its Nobel-prize winning AlphaFold team", "excerpt": "Google DeepMind disbands its Nobel-prize winning AlphaFold team https://www.engadget.com/2225849/google-shuts-down-alphafold/ https://news.ycombinator.com/item?id=49098375", "context": "", "use_case": "", "switched_from": "", "story": "Gemini Robotics 2 brings whole body intelligence to robots"}, {"id": 785626208, "item_id": 49101638, "url": "https://news.ycombinator.com/item?id=49101638", "by": "westurner", "date": "2026-07-29", "version": "3", "stance": "neutral", "quote": "AlphaFold 3 inference pipeline", "excerpt": "google-deepmind/alphafold3: AlphaFold 3 inference pipeline https://github.com/google-deepmind/alphafold3 And then host then database. AlphaFold did not yet solve the Folding@home protein folding problem. Folding@home: https://en.wikipedia.org/wiki/Folding@home", "context": "", "use_case": "", "switched_from": "", "story": "Google shuts down Nobel Prize winning AlphaFold"}, {"id": 777273664, "item_id": 48579604, "url": "https://news.ycombinator.com/item?id=48579604", "by": "astrange", "date": "2026-06-17", "version": null, "stance": "recommend", "quote": "I like how you're trying to argue with the Nobel committee here.", "excerpt": "https://www.the-scientist.com/chatgpt-and-alphafold-help-design-personalized-vaccine-for-dog-with-cancer-74227 https://arxiv.org/pdf/2503.15204 > They already fold proteins with simulations. I like how you're trying to argue with the Nobel committee here.", "context": "", "use_case": "", "switched_from": "", "story": "Leaked financial docs show OpenAI is losing billions of dollars a year"}, {"id": 774940288, "item_id": 48433768, "url": "https://news.ycombinator.com/item?id=48433768", "by": "a_bonobo", "date": "2026-06-07", "version": "2", "stance": "recommend", "quote": "Alphafold2 was great but didn't actually 'solve' protein-folding", "excerpt": "The accompanying preprint is interesting: https://www.biorxiv.org/content/10.64898/2026.06.03.729735v1 Modeling protein-protein binding is still a massively unsolved problem, mainly because we don't really have the data. Alphafold2 was great but didn't actually 'solve' protein-folding as all input data is from single 'state' X-ray crystallography of the proteins, not 'really' how these proteins behave in the wild. So", "context": "", "use_case": "", "switched_from": "", "story": "Biohub releases a world model of protein biology"}, {"id": 774205728, "item_id": 48387858, "url": "https://news.ycombinator.com/item?id=48387858", "by": "photochemsyn", "date": "2026-06-03", "version": null, "stance": "warn", "quote": "This does reveal the weakness of AlphaFold approaches", "excerpt": "This does reveal the weakness of AlphaFold approaches for answering questions like \u201cwhat is possible in the protein folding space if you use the 20 canonical amino acids\u201d since the data used to train AlphaFold is limited to existing experimentally determined protein structures. We don\u2019t even know if this is like body plans (four legs for mammals, why not six?) i.e. is this about physical limitations of the folding sp", "context": "", "use_case": "", "switched_from": "", "story": "The Unreasonable Redundancy of Nature's Protein Folds"}, {"id": 773854145, "item_id": 48365884, "url": "https://news.ycombinator.com/item?id=48365884", "by": "wyre", "date": "2026-06-01", "version": null, "stance": "recommend", "quote": "Literally saved his dog's life.", "excerpt": "https://www.the-scientist.com/chatgpt-and-alphafold-help-design-personalized-vaccine-for-dog-with-cancer-74227 Literally saved his dog's life.", "context": "", "use_case": "Dog cancer vaccine design", "switched_from": "", "story": "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"}, {"id": 772533298, "item_id": 48283331, "url": "https://news.ycombinator.com/item?id=48283331", "by": "Subtextofficial", "date": "2026-05-26", "version": null, "stance": "warn", "quote": "hours on AlphaFold queue", "excerpt": "Built this in a week on owned hardware (4x GPU, 32 shards). Running on a single machine in Portland with $126/mo overhead. Live endpoints: - Emotion detection: 16ms P50, 10x faster than Azure at same price - CRISPR guide RNA design: only self-serve API, Benchling charges $20k+/year - Protein folding: ESMFold, sub-10s vs hours on AlphaFold queue - Local LLM: 97% cheaper than OpenAI, data never leaves the server - Town", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 766741425, "item_id": 47921339, "url": "https://news.ycombinator.com/item?id=47921339", "by": "squid-protocol", "date": "2026-04-27", "version": null, "stance": "warn", "quote": "It revealed a very linear data pipeline stuffed with steps of O(N)^6 algorithms.", "excerpt": "I've been working on a custom AST-free, LLM-free code base analysis engine and used it to generate the following data to dissect Alphafold. It revealed a very linear data pipeline stuffed with steps of O(N)^6 algorithms. repo - https://github.com/squid-protocol/gitgalaxy", "context": "", "use_case": "", "switched_from": "", "story": "Teardown of AlphaFold codebase with a AST-free LLM-free static analysis engine"}, {"id": 754012432, "item_id": 47125777, "url": "https://news.ycombinator.com/item?id=47125777", "by": "drb493", "date": "2026-02-23", "version": null, "stance": "recommend", "quote": "AlphaFold is an amazing machine learning approach to predicting protein structure", "excerpt": "I'd discourage claiming any biological process is \"solved.\" But to your point: No--AlphaFold is an amazing machine learning approach to predicting protein structure but Folding@Home is still immensely useful for simulating how proteins fold up over a timescale. They are/will be complimentary methods. https://pmc.ncbi.nlm.nih.gov/articles/PMC11892350/", "context": "", "use_case": "Protein structure prediction", "switched_from": "", "story": "SETI@home: Data Acquisition and Front-End Processing"}, {"id": 754006176, "item_id": 47125386, "url": "https://news.ycombinator.com/item?id=47125386", "by": "alexpotato", "date": "2026-02-23", "version": null, "stance": "recommend", "quote": "Wasn't this largely solved by DeepMind's AlphaFold?", "excerpt": "Wasn't this largely solved by DeepMind's AlphaFold? https://alphafold.ebi.ac.uk/", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "SETI@home: Data Acquisition and Front-End Processing"}, {"id": 751374416, "item_id": 46960901, "url": "https://news.ycombinator.com/item?id=46960901", "by": "nycdatasci", "date": "2026-02-10", "version": "3", "stance": "compare", "quote": "more than doubles the accuracy of AlphaFold 3", "excerpt": "\"We demonstrate that our IsoDDE more than doubles the accuracy of AlphaFold 3 on a challenging protein-ligand structure prediction generalisation benchmark, predicts small molecule binding-affinities with accuracies that exceed gold-standard physics-based methods at a fraction of the time and cost, and is able to accurately identify novel binding pockets on target proteins using only the amino acid sequence as input.", "context": "", "use_case": "", "switched_from": "", "story": "The Isomorphic Labs Drug Design Engine unlocks a new frontier beyond AlphaFold"}, {"id": 750880128, "item_id": 46930008, "url": "https://news.ycombinator.com/item?id=46930008", "by": "tim333", "date": "2026-02-07", "version": "3", "stance": "recommend", "quote": "Maybe along the lines of alphafold 3", "excerpt": "Maybe along the lines of alphafold 3 https://blog.google/innovation-and-ai/products/google-deepmind-isomorphic-alphafold-3-ai-model/#alphafold-server", "context": "", "use_case": "", "switched_from": "", "story": "The AI boom is causing shortages everywhere else"}, {"id": 740351680, "item_id": 46271980, "url": "https://news.ycombinator.com/item?id=46271980", "by": "D-Machine", "date": "2025-12-15", "version": null, "stance": "recommend", "quote": "it is a highly unique architecture and approach", "excerpt": "This article is garbage and makes many incorrect claims, and it is clearly AI-generated. E.g. the claim that \"AlphaFold doesn't simulate physics. It recognizes patterns learned from 170,000+ known protein structures\" couldn't be farther from the truth. Physical models are baked right into AlphaFold models and development at multiple steps, it is a highly unique architecture and approach. AlphaFold models also used TP", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "Why proteins fold and how GPUs help us fold"}, {"id": 740351584, "item_id": 46271974, "url": "https://news.ycombinator.com/item?id=46271974", "by": "D-Machine", "date": "2025-12-15", "version": null, "stance": "recommend", "quote": "Physical models are baked right into AlphaFold models and development at multiple steps", "excerpt": "Yes https://github.com/google-deepmind/alphafold/issues/31#issuecomment-882623226. This article is garbage and makes many incorrect claims, and it is clearly AI-generated. E.g. the claim that \"AlphaFold doesn't simulate physics. It recognizes patterns learned from 170,000+ known protein structures\" couldn't be farther from the truth. Physical models are baked right into AlphaFold models and development at multiple st", "context": "", "use_case": "Protein structure prediction", "switched_from": "", "story": "Why proteins fold and how GPUs help us fold"}, {"id": 740350080, "item_id": 46271880, "url": "https://news.ycombinator.com/item?id=46271880", "by": "D-Machine", "date": "2025-12-15", "version": "2", "stance": "recommend", "quote": "See for example the AlphaFold2 presentation linked here", "excerpt": "See for example the AlphaFold2 presentation linked here: https://predictioncenter.org/casp14/doc/presentations/2020_12_01_TS_predictor_AlphaFold2.pdf. Some samples that point out where most of the innovations are NOT just \"huck a transformer at it\": ==== Physical insights are built into the network structure, not just a process around it - End-to-end system directly producing a structure instead of inter-residue dist", "context": "", "use_case": "protein structure prediction", "switched_from": "", "story": "Why proteins fold and how GPUs help us fold"}, {"id": 737908691, "item_id": 46119293, "url": "https://news.ycombinator.com/item?id=46119293", "by": "Glemkloksdjf", "date": "2025-12-02", "version": null, "stance": "recommend", "quote": "AlphaFold solved protein folding.", "excerpt": "Image editing is now working so well, last week i used nano banana instead of doing anything in photoshop. That image generation is already disrupting industries and jobs today. My mother! (non technical) had a call with a support ai agent just a few month ago. AI is also not a fad. LLMs are the best interface we had sofar. They will stay. AI Coding helps non developers to develop a lot faster and better (think resea", "context": "", "use_case": "", "switched_from": "", "story": "It\u2019s been a very hard year"}, {"id": 737622160, "item_id": 46101385, "url": "https://news.ycombinator.com/item?id=46101385", "by": "theturtletalks", "date": "2025-11-30", "version": null, "stance": "recommend", "quote": "", "excerpt": "Didn\u2019t the transformer model come from AlphaFold? I feel like we wouldn\u2019t have had the LLMs we use today if it wasn\u2019t for AlphaFold.", "context": "", "use_case": "", "switched_from": "", "story": "The Thinking Game Film \u2013 Google DeepMind documentary"}]}