CASP, The Contest That Inspired AlphaFold Scrambles for Survival After 30 Years of Advancing AI in Biology as NIH Funding Dries Up and Google Remains Silent
New Open-Source IntFOLD Model Served Up by Tamarind Bio Rivals AlphaFold 3 in Antibody-Antigen Complex Prediction
Unnatural Products Scores Largest-Yet Macrocycle Licensing Deal with $1.5B Argenx Partnership
Not yet a member of our super awesome slack community of ~10,000? Join HERE 🤗
CASP's $800K Grant Expires in Weeks, Threatening Future of Competition That Revolutionized Structural Biology
The Critical Assessment of protein Structure Prediction (CASP), the 30-year-old competition credited with spurring AI development in biology including AlphaFold, has exhausted its NIH funding and will deplete emergency support from UC Davis by August 8, with the university issuing 45-day termination notices to program staff. Despite submitting an $800,000 grant renewal request last year, CASP organizers have received no response from NIH, which has been slow to spend its 2025 budget under the Trump administration, stoking fears that much of the agency's $48 billion allocation will go unused. CASP co-founder John Moult says organizers are "scrambling" to find alternative funding from foundations and other countries, while UC Davis explores whether it can afford to archive CASP's computer servers to preserve decades of resources. The competition's impact on structural biology has been transformative: in CASP1 (1994), predictions for challenging proteins were "near random," but by CASP16 last year, modelers achieved nearly 95% success rates, with researchers concluding "the problem of folding single proteins is nearly solved." Industry observers including MIT's Sergey Ovchinnikov have called for private support, with suggestions that DeepMind/Google should "donate generously" given CASP's role in AlphaFold's development and the subsequent revenues it has achieved. Even if funding materializes, CASP director Krzysztof Fidelis warns "it will not be easy to reconstruct" the program after staff departures and the disbanding of its scientific volunteer network.
Y Combinator-Backed Tamarind Launches IntFOLD to Democratize State-of-the-Art Protein Structure Prediction
While the pace of innovation in computational biology is breathtaking, many of the most powerful open-source tools remain out of reach for bench scientists without dedicated DevOps and GPU resources. Tamarind.bio is one of the companies aiming to solve this "deployment gap" with a no-code platform that makes state-of-the-art models accessible through a simple web interface. However, they dont stop there. Tamarind Bio has launched IntFOLD on its low/no code platform, a new protein structure prediction protocol that achieves performance comparable to AlphaFold 3 on the comprehensive FoldBench benchmark, with 0.88 LDDT scores for protein monomers and 72.9% success rate for protein-protein interactions. Developed by researchers at the University of Reading and Fudan University, the most recent version of IntFOLD+ shows particular strength in antibody-antigen complexes—a commercially critical area where alternatives have lagged—achieving 43.2% success rate compared to AlphaFold 3's 47.9%. When incorporating known binding site constraints, IntFOLD demonstrates dramatic improvements: the PoseBusters dataset success rate jumps from 79.5% to 89.7%, while antibody-antigen interface predictions soar from 37.6% to 69.0%. The model features technical advances including a custom FlashAttentionPairBias kernel for improved efficiency and a model-agnostic ranking method that boosts success rates by ~3% through structural similarity consensus. Tamarind, founded by Stanford CS graduates Deniz Kavi and Sherry Liu, has already attracted over >1,000 scientists from institutions all over the world Stanford, Harvard, and Oxford to its platform, which offers no-code access to computational biology tools at scale. This release positions Tamarind as a serious competitor in the protein structure prediction space, offering pharmaceutical companies a fully commercial alternative.
Argenx commits up to $1.5 billion for Unnatural Products' unique oral macrocyclic peptides
Netherlands-based Argenx has entered a multi-target collaboration potentially worth $1.5 billion with Unnatural Products (UNP) to develop oral macrocyclic peptides for "undruggable" disease targets, involving a double-digit million-dollar upfront payment plus research funding, milestones, and tiered royalties. UNP's platform, touted as "the first scalable platform to engineer synthetic macrocycles using AI-guided design, parallel synthesis, and direct-to-biology screening," will generate drug candidates through IND-enabling studies before argenx takes over clinical development. The timing is strategic for argenx, whose business currently depends heavily on the FcRn blocker efgartigimod (Vyvgart/Vyvgart Hytrulo), which brought in $2.2 billion in 2024 but faces future patent risks. Unlike traditional antibodies that require infusion or injection, UNP's macrocyclic peptides can be delivered as simple pills with shelf stability, providing more surface area to disrupt protein-protein interactions while resisting breakdown by proteases. According to a UNP spokesperson, this represents the largest licensing deal to date in the macrocyclic peptide space, with argenx also pledging to participate in UNP's upcoming Series B round. The roughly 40-person UNP team, which previously secured deals with BridgeBio for rare fibrotic diseases and Merck & Co. for oncology targets, continues advancing its internal pipeline alongside partnered assets. This partnership is more than just a large number; it's a landmark deal in one of the hottest therapeutic modalities and signals a major strategic pivot by a leading immunology company and a massive vote of confidence in an AI-driven discovery platform.
Owkin secures five-year partnership with Newcastle NHS to develop AI that screens for breast cancer mutations directly from H&E slides
Agentic AI company Owkin and Newcastle upon Tyne Hospitals NHS Foundation Trust have entered a five-year partnership to explore how agentic AI can accelerate drug discovery, development, and diagnostics, with Newcastle supporting development of BRCAura®, an AI-powered product that screens for gBRCA mutations in breast cancer directly from digitized pathology slides. By leveraging routinely generated H&E slides rather than requiring separate genetic testing, BRCAura® could dramatically streamline the gBRCA testing process, allowing oncologists and genetic counselors to accelerate testing for high-risk patients and incorporate results into treatment planning without delays. Newcastle joins Owkin's ATLANTIS program, an international initiative of 20 leading hospitals across seven countries working to optimize and harmonize patient data for AI-driven research. The partnership is particularly significant given Newcastle's status as one of the UK's largest NHS trusts and home to the Northern Centre for Cancer Care (NCCC), which provides cutting-edge radiotherapy, chemotherapy, immunotherapy, and surgical oncology services. This follows Owkin's track record with UK institutions, including work with Guy's and St Thomas' NHS Foundation Trust on HER2-positive breast cancer detection, where up to 14,000 women in Britain are affected annually. With Sanofi's $180 million investment making Owkin a unicorn in 2021, the company continues to expand its federated learning network that enables data scientists to train AI models on decentralized datasets while preserving patient privacy.
PathAI scores FDA clearance for AISight Dx digital pathology platform with built-in approval pathway for future AI algorithms
Summary: One of the original histopathology AI companies, PathAI has received a much awaited 510(k) clearance from the FDA for AISight Dx, its cloud-native digital pathology image management system for primary diagnosis, building on the initial clearance for AISight Dx(Novo) in 2022. Critically, the FDA included a Predetermined Change Control Plan (PCCP), enabling PathAI to validate and implement major changes such as additional displays, scanners, file formats, and browsers without requiring new 510(k) submissions—a regulatory first that could accelerate AI deployment in pathology. The platform features intelligent case management, synchronized multi-slide navigation, and real-time collaboration capabilities, allowing pathologists to work remotely and conduct expert consults without physical specimen review. AISight Dx is FDA-cleared for use with Hamamatsu NanoZoomer S360MD and Leica Aperio GT 450 DX scanners, and also carries CE-IVD marking for primary diagnosis in the EEA, UK, and Switzerland. The clearance comes as PathAI launches its Precision Pathology Network, a first-of-its-kind network of digital anatomic pathology laboratories that will provide early access to AI-powered solutions developed with biopharma partners and enable labs to monetize real-world pathology data.
Circulate Health lands $12 million from Khosla Ventures to expand plasma exchange network after human trial demonstrates 2.61-year biological age reduction
A spin out from the Buck Institute, Circulate Health has raised $12 million in seed funding led by Khosla Ventures to scale its therapeutic plasma exchange (TPE) service, which removes inflammatory factors and age-related compounds from blood, following a human trial showing participants who received TPE with immunoglobulin reduced their biological age by 2.61 years on average. The company's patented procedure separates, removes, and replaces plasma to treat inflammation and improve healthspan, building on decades of medical use for blood disorders, autoimmune diseases, and neurological conditions. Co-founded by former 98point6 CMO Dr. Brad Younggren and Buck Institute CEO Dr. Eric Verdin, Circulate already operates in 24 partner clinics across 8 states and expects to expand to over 50 clinics by the end of 2025. The 42-person study published in Aging Cell compared three groups: placebo, TPE alone (1.32-year reduction), and TPE with antibodies (2.61-year reduction), with participants showing improvements in biomarkers consistent with cellular rejuvenation. Alex Morgan of Khosla Ventures noted the company was incubated to explore whether TPE could counter aging's deleterious effects and potentially remove microplastics from the body. At around $36,000 per treatment series according to industry reports, the service joins a growing longevity tech sector that attracted $1.8 billion in investments from 2021 through early 2025, though Circulate emphasizes its FDA-cleared approach differs from unproven "biohacking" treatments.
French Startup Rainpath Raises €2.5M to Replace Toxic Chemical Stains with AI in Pathology Labs
Rainpath, founded just three months ago by engineers Mel Landelle and Antoine Delplace, has raised €2.5 million in seed funding led by Teampact Ventures and regional fund Xplore by Épopée Gestion to revolutionize biopsy analysis by replacing traditional chemical staining with artificial intelligence. The technology promises to divide analysis time and costs by 100 while completely eliminating the use of toxic chemicals like hematoxylin-eosin, addressing critical bottlenecks in anatomopathology labs including extended wait times, exploding reagent costs, and recurring manual errors. The round also includes The Quest, Advans Lab, Sharpstone, fifteen business angels, and support from Bpifrance, with Teampact Ventures highlighting the dual revolution of enabling tele-anatomopathology while eliminating toxic waste—perfectly aligned with their Climate & Health strategy. The 25-year-old CEO Landelle, a graduate of École normale supérieure and École polytechnique's Master PIC, aims to position Rainpath as "the Cardiologs of anatomopathology," targeting a global market of over 500 million analyses performed annually. The company plans to double its team from 8 to 16 employees, recruiting R&D talent including AI engineers, data scientists, and a Chief Product Officer, while targeting Series A funding within 18 months for international expansion. This positions Rainpath at the forefront of France's AI-driven healthtech ecosystem, where diagnostic assistance startups like Gleamer, Milvue, and Tribun Health have already gained traction.
Centivax locks down $45 million to push first universal flu vaccine into human trials by Q2 2026
South San Francisco-based Centivax has closed an oversubscribed $45 million Series A led by Steve Jurvetson's Future Ventures to advance its universal flu vaccine candidate through the final steps of preclinical development and into Phase I trials slated to begin within eight months. The company's counterintuitive approach encodes 22 highly diverse flu proteins from the past century via mRNA, each at sub-threshold doses, forcing the immune system to target conserved regions that don't mutate between strains. The Phase I study will use the gold-standard hemagglutination inhibition (HAI) assay to measure efficacy against more than twenty flu strains—including current circulating strains, historical mismatch strains, and pandemic strains—in direct comparison with existing vaccines. Beyond influenza, Centivax's “epitope-focusing” platform has shown promise for RSV/hMPV, herpesviruses, HIV, malaria, oncology, and even a universal antivenom recently published in Cell. With $24 million in previous non-dilutive funding from the Gates Foundation, CEPI, NIH, and military agencies, plus a leadership team including former Merck vaccine developer Dr. Jerald Sadoff (who helped develop 14 approved vaccines including Gardasil), Centivax aims to capture significant share of the $7 billion annual flu vaccine market. CEO and Netflix star Dr. Jacob Glanville projects FDA approval by 2029, aligning with the agency's own forecast for universal flu vaccine readiness.
Chugai commits up to $250 million for Gero's AI-discovered aging targets in pursuit of antibody therapies
Singapore-based Gero and Roche's Chugai Pharmaceuticals have entered a joint research and license agreement where Chugai will develop antibody drug candidates using its proprietary engineering technologies for targets discovered by Gero's AI platform, which analyzes human datasets to identify drivers of age-related diseases. Under the agreement, Gero grants Chugai exclusive worldwide rights for the creation, research, development, manufacturing, and commercialization of antibodies for the identified targets, with Chugai paying an undisclosed upfront plus up to approximately $250 million in development and sales milestones. According to a Gero representative, royalties could push the total deal value above $1 billion if products successfully launch. Gero CEO Peter Fedichev noted their AI platform is "built to identify therapeutic targets that drive multiple age-related diseases and potentially aging itself," aiming to translate insights into therapeutics that restore lost function. The partnership aligns with Chugai's “TOP I 2030” growth strategy emphasizing open innovation with global partners, as articulated by President and CEO Dr. Osamu Okuda. Gero, which raised a $6 million Series A in 2023 and previously partnered with an undisclosed pharma company on fibrotic diseases, remains in early stages with no published drug pipeline, making this Roche subsidiary partnership a significant validation of its physics-based machine learning approach to aging biology.
See our handy dandy Lu.ma event calendar HERE, please RSVP so folks can plan accordingly!
Supply Chain Logistics in Biopharma, Thursday, July 17th in San Francisco
Thank you to Eti Sinha and Shantenu Agarwal for organizing!
Minneapolis-St Paul Get Together, Thursday, July 24th
Thank you to Brian Repko for organizing!
@William Steele has written a tutorial for using a very cool, albeit highly specialized, database of antibodies
Postdoctoral Fellow at Université Paris-Saclay
Product Manager at Deep Origin
Founding Engineer at Axia Medicine
Feedback: How is the Newsletter doing? We’re trying different formats/content. In case the hyperlink above didn’t get your attention, maybe a bright orange button will!
Volunteer: Want to get involved with Bits in Bio, meet new members across the community, and learn about the ecosystem? We are looking for volunteers to help us create great content and manage the community.
We want to deliver what matters most in Bio AI and would love your feedback on how we can do better. Please weigh in as anon here or DM me directly!
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.