Protein design
4 labs · 9 models and products Theme: Companies · Generated from this page's data. Click a name to open its panel on the Labs page. Records resting on one source are tagged "one source".Proteins do most of the work in living cells, and their shape decides what they do. Protein design uses AI to invent new proteins, for example antibodies or small "binders" that latch onto a disease target, instead of finding them in nature. The 2024 Nobel Prize in Chemistry went half to David Baker for computational protein design and half to Demis Hassabis and John Jumper, AlphaFold's creators, for structure prediction. Latent Labs has published two design models, Latent-X and Latent-X2; Antiverse, in Cardiff, uses AI to design antibody drugs (per the site).
Context
- Demis Hassabis and John Jumper shared half of the 2024 Nobel Prize in Chemistry; David Baker received the other half for computational protein design. 2024 (page undated; accessed 31 October 2026) nobelprize.org/prizes/chemistry/2024/popular-inf
for protein structure prediction
- The Nobel committee says AlphaFold2 can predict almost all known protein structures and had been used by over two million people by October 2024. October 2024 nobelprize.org/prizes/chemistry/2024/popular-inf
more than two million people from 190 countries
- Google DeepMind's AlphaProteo designs new proteins that bind tightly to chosen targets. 12 September 2024 arxiv.org/abs/2409.08022
This technical report introduces AlphaProteo, a family of machine learning models for protein design
- Latent Labs reports lab hit rates above 90% for ring-shaped peptides designed with Latent-X. 25 July 2025 arxiv.org/abs/2507.19375
For macrocyclic peptides, Latent-X achieves experimental hit rates exceeding 90% on all evaluated benchmark targets.
- Latent Labs reports Latent-X2 antibodies with drug-like properties matching or beating approved antibody medicines. 23 December 2025 arxiv.org/abs/2512.20263
Designed molecules exhibit developability profiles that match or exceed those of approved antibody therapeutics
Key papers
- De novo design of high-affinity protein binders with AlphaProteo arXiv preprint · 12 September 2024
- Latent-X: An Atom-level Frontier Model for De Novo Protein Binder Design arXiv preprint · 25 July 2025
- Drug-like antibodies with low immunogenicity in human panels designed with Latent-X2 arXiv preprint · 23 December 2025
Labs Round 8 (31 Oct 2026): introduction and facts as the researchers wrote them; "(per the site)" marks a sentence resting on this site's own description of that lab.
| Record | What it is | Place | Links | Confidence |
|---|---|---|---|---|
| AbBFN2 (InstaDeep)Chips, models and products | InstaDeep's generative model for designing and humanising antibodies. announcedKey facts Bayesian Flow Network; InstaDeep reports 90% humanisation success with tractable candidates in under 20 minutes (2025-10-13)Made by InstaDeep · 2025-10-13 | London (NW1 3BF ) (via InstaDeep) | instadeep.com/2025/10/ai-day-2025-powering-biolo | one source |
| AlphaProteoChips, models and products | An AI system that designs new proteins that bind tightly to chosen target molecules. released (paper only)Key facts Theme: Protein design. Announced 5 Sep 2024; whitepaper on arXiv; phased sharing, no public tool stated. Figures (5 Sep 2024): 3 to 300 times better binding affinity than best existing methods on seven targets; 88% of candidates bound to BHRF1 in lab tests.Made by Google DeepMind · 2024-09-05Read the paper | London (EC2M 4AG) (via Google DeepMind) | deepmind.google/blog/alphaproteo-generates-novel · deepmind.google/blog/page/8/ | one source |
| AntiConf (antibody-antigen confidence score)Chips, models and products | Antiverse method that scores how far to trust AI-predicted structures of antibodies bound to their targets. released (paper)Key facts Published in Briefings in Bioinformatics, dated 28 Mar 2026 on the Antiverse platform page; access: paperMade by Antiverse · 2026-03-28 | Cardiff (CF24 4HQ) (via Antiverse) | antiverse.io/platform | one source |
| AntiverseCompanies · also on Start-ups | Cardiff company that uses AI to design antibody drugs against targets that are hard to reach with existing methods.Profile
Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: verified (Series A: company news page plus trade); total raised one source. Sources Models and products (2)
Key facts and news (5)
Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked. | Cardiff (CF24 4HQ) | Companies House · uktech.news/ai/welsh-biotech-raises-7m-for-ai-de · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov. · uktech.news/featured/wales-antibody-drug-discove | verified |
| Antiverse PlatformChips, models and products | Antiverse's AI system for designing antibodies against hard drug targets such as GPCRs and ion channels. live (in-house and partnered discovery)Key facts Generative de novo design plus epitope-specific function prediction and developability filters; company-stated figures: 2000x greater hit rate than a naive library; 100 billion antibody-antigen interactions probed; more than 240 proprietary stable cell lines; GPCR data collected since 2017; access: partnershipsMade by Antiverse · — (date not stated) | Cardiff (CF24 4HQ) (via Antiverse) | antiverse.io/platform · antiverse.io/ | one source |
| Latent LabsCompanies | Builds AI models and an AI agent that design new proteins, such as antibodies, for drug and biotechnology companies.Profile
Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source. Sources Models and products (3)Version chainsLatent-X (Jul 2025) → Latent-X2 (Dec 2025) → Latent-Y agent (Mar 2026)
Selected papers (3)
Key facts and news (3)
News
Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked. | White Collar Factory, Old Street Yard, London (EC1Y 8AF) | Companies House · latentlabs.com/press-release/latent-labs-secures · tech.eu/2025/02/13/ai-pioneer-dr-simon-kohls-lat · latentlabs.com/press-release/latent-y/ · latentlabs.com/press-release/ | verified |
| Latent-XChips, models and products | Latent Labs' first public AI model, which designs small protein binders and ring-shaped peptides ready for lab testing. released (no-code web platform, free tier)Key facts Launched 22 Jul 2025 on a browser-based platform with a free tier for commercial and non-commercial users; technical report (arXiv 2507.19375). Figures (22 Jul 2025): macrocycle hit rates 91–100%; mini-binder hit rates 10–64%; picomolar mini-binder affinities; tested on 7 therapeutic targets. Version history: Latent-X (Jul 2025) → Latent-X2 (Dec 2025) → Latent-Y agent (Mar 2026).Made by Latent Labs · 2025-07-22Versions Latent-X (Jul 2025) → Latent-X2 (Dec 2025) → Latent-Y agent (Mar 2026)Read the paper | White Collar Factory, Old Street Yard, London (EC1Y 8AF) (via Latent Labs) | latentlabs.com/press-release/introducing-latent- · latentlabs.com/press-release/ | lab's own pages |
| Latent-X2Chips, models and products | Latent Labs' second model, which designs antibodies and ring-shaped peptides with properties suited to drug development. early access (web and API; commercial use by partnership)Key facts Announced 16 Dec 2025; technical report arXiv 2512.20263. Figures (16 Dec 2025): lab-validated antibody hits from 4 to 24 designs per target; confirmed hits on half of 18 targets; low immunogenicity in a ten-donor human panel; scFv binders at low picomolar affinity.Made by Latent Labs · 2025-12-16Read the paper | White Collar Factory, Old Street Yard, London (EC1Y 8AF) (via Latent Labs) | latentlabs.com/latent-x2/ · latentlabs.com/press-release/ | lab's own pages |
| Latent-YChips, models and products | An AI agent that designs antibodies from written goals, running Latent-X2 inside an automated workflow. released (selected partners; Research Tier for researchers)Key facts Announced 23 Mar 2026. Figures (23 Mar 2026): design campaigns 56-fold faster in user studies; 67% target-level success across nine targets; lab-confirmed binders on six targets at single-digit nanomolar affinity. Research Tier launch event announced 23 Jun 2026; launched to researchers worldwide with free daily access 15 Jul 2026 (index title; release page not opened).Made by Latent Labs · 2026-03-23Read the paper | White Collar Factory, Old Street Yard, London (EC1Y 8AF) (via Latent Labs) | latentlabs.com/latent-y/ · latentlabs.com/press-release/ | lab's own pages |
| PeptoneCompanies · also on Start-ups | Protein-engineering software company that uses AI and physics-based computing to help design medicines.Models and products (1)
Selected papers (2)
Key facts and news (1)
Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked. | (registered office NG18 1RR) | Companies House · uktech.news/news/uk-biotech-peptone-funding-for- · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov. · peptone.io/ | verified |
| PepTron-o (Peptone)Chips, models and products | Peptone and NVIDIA model that predicts the shifting shapes of "disordered" proteins. announcedKey facts Announced with NVIDIA 2025-06-12Made by Peptone · 2025-06-12 | (registered office NG18 1RR) (via Peptone) | peptone.io/news/ | one source |
| SynthID BioChips, models and products | A proof of concept for watermarking AI-designed proteins without changing how they work. proof of conceptKey facts Theme: Protein design. Announced 30 Sep 2026.Made by Google DeepMind · 2026-09-30 | London (EC2M 4AG) (via Google DeepMind) | deepmind.google/blog/rss.xml | one source |
| XymeCompanies | Oxford company founded in 2024 by Exscientia founder Andrew Hopkins that designs new enzymes with AI and quantum chemistry.Key facts Founded 2024; combines AI, quantum chemistry and lab experiments; took 5,600 sq ft at Inventa, Oxford West End (reported 2025-05-30) | Oxford | Companies House · cambridgenetwork.co.uk/index.php/news/mission-st · find-and-update.company-information.service.gov. | one source |