Bits In Bio Weekly - April 22nd 2025
Updates From and Around the Bits in Bio Community
📢 Highlights
R&D Arms Race - Models Show China Leaving the U.S. 30 %–60 % Behind
Sanofi Bets $1.8 B on Earendil’s AI-Engineered Bispecifics for IBD
PacBio and Sanger Team Up to Decode Immunity One Long-Read RNA Sequence at a Time
Not yet a member of our super awesome slack community of >8500? Join HERE 🤗
👀 In Case You Missed it …
Study: China to eclipse U.S. in R&D spending by 2030
New analyses suggest that by the end of this decade, China will be outspending the United States on R&D by than 30-60%, even under pessimistic scenarios of continued trade wars. Using OECD data and 1,000 simulated models, researchers found the outcome was the same every time: China’s research investment grows relentlessly and overtakes the U.S. around 2030 in purchasing power terms – and keeps widening the gap. This projection holds true even if high tariffs and tech export restrictions persist (which could slow China a bit, but not enough to flip the trend). In practical terms, China is pouring money into science and technology at an unprecedented scale, while U.S. federal R&D funding has been comparatively stagnant. The implications are sobering for American innovation leadership; experts are calling it a “Sputnik moment” that should spur the U.S. to reinvest heavily in science. Otherwise, we’re looking at a world where China could dominate key research fields (from AI to biotech) through sheer scale of funding. The report underscores that U.S. policy decisions (like cutting R&D budgets) will have outsized impact on this race – and that maintaining a competitive edge may require a serious boost in public research investment to keep pace with China’s trajectory.
Behind the Deal: Sanofi’s $1.8B Leap into AI-Engineered Immunology
Sanofi has entered into a significant licensing agreement with Earendil Labs, a Delaware-based biotech firm specializing in AI-driven biologics, to acquire exclusive global rights to two bispecific antibody candidates: HXN-1002 and HXN-1003. The deal includes an upfront payment of $125 million, with potential milestone payments totaling up to $1.72 billion, and tiered royalties on future sales.
Earendil’s proprietary AI platform integrates advanced machine learning, generative protein engineering, and high-throughput experimental techniques to design and optimize protein-based therapeutics. This approach enables the rapid identification of antibody candidates with enhanced functionality and developability.HXN-1002 targets α4β7 and TL1A, aiming to treat moderate to severe ulcerative colitis and Crohn’s disease. HXN-1003 targets TL1A and IL23, addressing colitis and skin inflammation. The market for inflammatory bowel disease (IBD) therapies is huge, and unmet needs are high because many patients fail to achieve sustained remission despite existing treatments, leading to persistent symptoms and complications. This collaboration is consistent with Sanofi’s multi-year strategic focus on expanding its immunology pipeline, becoming “an immunoscience powerhouse” and leveraging AI technologies to accelerate drug discovery at every step along the way.
Sanger Institute and PacBio launch long-read, single-cell immune cell atlas
A new initiative out of the UK is harnessing long-read sequencing at single-cell scale to map the “hidden” complexity of our immune system. The Wellcome Sanger Institute, in partnership with Pacific Biosciences, will sequence about 1,500 blood and gut samples with long-read single-cell RNA-seq, capturing not just which genes are active in individual immune cells, but the exact isoforms (gene variants) being used. This is groundbreaking because long reads can span entire transcripts, revealing alternative splicing and gene versions that short-read methods miss. The project focuses on three studies: two in inflammatory bowel disease (IBD) – aiming to find RNA isoforms linked to IBD severity or treatment response – and one (Project JAGUAR) examining how genetic ancestry affects immune responses by including samples from Peru and Mexico. It’s the first time Sanger will deploy long-read sequencing at this scale, and they plan to openly share the high-resolution atlas of immune gene activity that results. Besides new biological insights (like potential isoform targets for IBD therapy), the team will also develop computational methods to handle the deluge of long-read data, hoping to lower the barriers for others to use these cutting-edge techniques. In short, this collaboration is pushing the frontier of genomics to understand immunity in unprecedented detail – down to how a single immune cell’s RNA may differ from those around it.
Recursion taps real-world data to “industrialize” clinical trials
Salt Lake City based Recursion is integrating de-identified health data from 340 million patients into its AI-powered platform to turbocharge clinical trial design. Through a new partnership, HealthVerity’s vast real-world dataset will feed Recursion’s machine learning “OS,” enabling smarter patient stratification, feasibility analyses, and streamlined trial operations. The goal is to make clinical development more like an industrial process – cutting costs and timelines by using real-world evidence to inform decisions at every step. Recursion’s leadership says coupling AI with nationwide patient data could unlock insights that improve trial success rates and speed novel therapies to market.
Potato AI raises $4.5M to automate science with AI and automation
A Seattle startup named Potato has landed $4.5 million in funding to advance its mission: fully automating the scientific experimentation process. Potato’s platform acts as an AI-powered research assistant that can help scientists generate hypotheses, design optimized experiments, and even execute them using robotic lab setups. Co-founders Ryan Kosai and Nick Edwards describe it as letting researchers input a question and having Potato sift through literature, plan out the best experiment (controls, reagents, protocols), and then send instructions to automated lab equipment to run it – essentially a self-driving laboratory. The idea is to dramatically speed up research by removing mundane manual steps and letting AI brainstorm and test ideas quickly. Early traction came from academic labs looking to boost throughput, and now with seed funding led by local VCs, Potato plans to refine its AI models and expand its fleet of robotic instruments. It’s part of a broader “lab of the future” trend. Keep an eye on this space: today it’s automating simple workflows, tomorrow it could be tackling complex multi-step discovery processes with minimal human intervention.
GeneDx acquires Fabric Genomics to scale AI-driven genetic testing
Diagnostic genomics company GeneDx (formerly part of Sema4) is buying Fabric Genomics, an AI genomic interpretation firm, in a deal worth up to $51M. The marriage will combine GeneDx’s huge rare disease sequence database (they’ve amassed over 750,000 exomes/genomes) with Fabric’s cloud-based software that uses AI to help interpret genetic variants. The strategic aim is to enable “decentralized” genome interpretation: hospitals around the world could sequence patients locally but then leverage GeneDx/Fabric’s platform as a service to analyze the data and identify disease-causing mutations. A curious definition. This model allows for global reach while ensuring consistent, expert analysis – and it addresses data sovereignty concerns by letting data stay in-country with just insights shared. Key use cases include rapid NICU sequencing (diagnosing critically ill newborns), newborn screening programs, and scaling clinical genomics in regions that lack specialized geneticists. Fabric’s platform is already deployed in major health systems, so with GeneDx’s resources it can now ramp up. Interestingly, Fabric will continue operating somewhat independently post-merger, but with GeneDx’s backing to expand. GeneDx is ultimately positioning itself for the next era of genomic medicine, where interpretation is the bottleneck, not sequencing.
ELIX and PRISM team up: AI + peptide mimics target the “undruggable” protein interactions
Two startups are teaming up to drug targets long deemed “undruggable.” Elix (an AI-guided drug design firm) and PRISM BioLab (which makes peptide-mimicking chemical scaffolds) announced a collaboration to tackle protein–protein interactions (PPIs) – a class of targets notoriously tricky for small molecules. By combining Elix’s AI compound-design platform with PRISM’s library of molecular scaffolds that imitate protein binding motifs, the partners plan to generate novel drug candidates for PPIs implicated in cancer, fibrosis, and autoimmune diseases. An earlier pilot already yielded distinct hits using this AI-plus-chemistry approach, hinting that tapping previously unexplored chemical space could finally crack PPI targets. The project aims to speed up discovery and improve success rates by iterating designs in silico and in the lab, potentially opening the door to therapies for diseases where conventional methods fell short.
Imaging giant RadNet acquires iCAD for AI breast cancer screening
One of the largest operators of imaging centers, RadNet, is acquiring iCAD, a pioneering maker of AI software for mammography. The $103M all-stock deal will fold iCAD’s ProFound AI suite for breast cancer detection into RadNet’s portfolio - RadNet also has its own DeepHealth AI tools. By combining forces, RadNet projects they’ll be affecting 10 million mammograms a year globally with their AI – essentially creating the industry’s most comprehensive set of AI solutions for breast imaging. For context, RadNet already handles ~4% of all US mammograms; adding iCAD’s tech and customer base extends reach to 1,700 sites in 50 countries. The rationale is clear: scaling up AI deployment can improve early detection and workflow efficiency in breast cancer screening, a field where every percentage gain in accuracy can save lives. RadNet’s CEO noted an interesting side benefit too: this move signals their commitment to digital health and has even attracted interest from hospital systems looking for partners in outpatient imaging. In short, it’s a big consolidation play in the radiology AI space – pairing data + distribution (RadNet’s clinics) with proven algorithms (iCAD’s decades of expertise) to hopefully catch breast cancers earlier and streamline diagnosis.
AI “bites” back at malaria in Venezuela’s gold rush with AI detection
Malaria is resurging in Venezuela due to illegal gold mining, but an AI tool is stepping in to help fight the outbreak. Researchers have developed a convolutional neural network that can automatically detect malaria parasites in blood smear images with 99.5% accuracy. They trained the model on nearly 190,000 augmented images, starting from about 6,000 microscope slides from Bangladesh, to teach it what the malaria parasites look like. Running on NVIDIA GPUs, the AI can analyze blood samples from remote mining regions where human experts and microscopes are scarce, flagging infected patients quickly. Notably, the work was just published in Nature, underscoring its scientific rigor. Venezuela was once declared malaria-free, so this tech arrives as the country grapples with a public health backslide. By deploying an AI-powered digital microscope in the field, health workers aim to catch infections early and accurately, despite limited medical infrastructure – a prime example of AI for social good in global health.
Ono bets on generative AI for RNA-editing therapies
The 300 year old Japanese titan, Ono Pharmaceuticals, has struck a deal with Cambridge, MA-based Jorna Therapeutics to design new RNA-editing drugs using generative AI. Jorna’s platform, intriguingly called “SkyEngine,” combines large-scale protein and RNA language models with quantum chemistry to craft molecules that can edit RNA – potentially fixing disease-causing RNA errors in patients’ cells. Under the collaboration, Jorna will use AI to generate RNA-editing sequences (likely guided by Ono’s target needs), and Ono secures exclusive worldwide rights to any drug candidates that come out. The financials weren’t fully disclosed, but include upfront payment, research funding, and milestones to Jorna – a sign Ono values this tech. Ono’s R&D leadership praised Jorna’s unique approach that designs bespoke RNA-editing proteins by analyzing tons of amino acid sequence data (think customizing enzymes to tweak RNA). The partnership aims to accelerate a new class of therapeutics for diseases with unmet needs, leveraging AI to conquer “previously inaccessible” RNA targets. It’s another example of pharma embracing AI-based design, this time in the cutting-edge realm of RNA editing (akin to ADAR or CRISPR-like editing on RNA).
Illumina + Tempus unite genomic data and AI for precision medicine
The DNA sequencing giant Ilumina is partnering with Tempus to propel genomic medicine into everyday healthcare. The collaboration will meld Illumina’s latest AI-driven analysis tools with Tempus’s enormous multimodal dataset (which includes genomic info plus clinical records, imaging, etc.) to train next-gen genomic algorithms. In practice, this means using real-world patient data to prove where genomic testing can impact care – not just in cancer (where it’s already used), but in cardiology, neurology, immunology, and beyond. Illumina’s vision is that in the near future, every complex disease patient gets routed to the optimal therapy based on molecular insights. A big barrier has been patchy evidence of benefit outside oncology; this team-up aims to generate that evidence by crunching Tempus’s data with AI and demonstrating improved outcomes with sequencing. In short, Illumina and Tempus want to make genomic profiling standard of care across all disease areas, by showing doctors and payers hard data on its value. It’s a notable alliance between a leading sequencer maker and a leading clinical data/AI company, highlighting the push to bring AI-powered precision medicine to the masses.
Johnson & Johnson trims its 900+ AI projects to 60
Even a powerhouse like J&J can fall victim to “shiny object syndrome” with AI – but now they’re refocusing. The company’s Chief Information Officer announced a pivot in J&J’s AI strategy: after encouraging nearly 900 experimental AI projects (“let a thousand flowers bloom”), they found only about 10-15% of those pilots delivered 80% of the value. So J&J is consolidating, doubling down on the proven winners and shelving the rest. In practice, that means channeling resources to high-impact use cases – e.g. AI tools that alleviate clinician burnout by automating routine documentation, or supply chain algorithms that meaningfully cut costs – and stopping low-value experiments that might be better solved with other tech. This more disciplined approach is part of a broader industry trend as the AI hype matures into reality: big healthcare companies want to see tangible ROI, not just cool demos. J&J’s move also sends a signal that quality beats quantity in AI initiatives. After a learning phase of wide-open experimentation, they’re now all about scale-up and implementation of the AI projects that actually move the needle for the business.
🗓️ Upcoming BiB Events
See our handy dandy Lu.ma event calendar HERE, please RSVP so folks can plan accordingly!
Bits in Bio April Journal Club - Drug Repositioning, Wednesday, April 23 at 5:30 PM PT
Thank you to Steve Cheney for organizing!
SynBioBeta BiB Happy Hour with Trinet, Wednesday, May 7 at 5:00 PM
Thank you to organizers Shantenu Agarwal, Beko Jang, Eti Sinha, and Poonam Lata
Chicago Bits in Bio Happy Hour, Tuesday, May 29 at 6:00 PM
First event in Chicago!
Thank you to Matthew Schechter for organizing
📰 Top Community Conversations
@Shantenu shows that the FDA’s shift to non-animal testing continues to drive dialogue
🏢 New Job Openings
Business Development Associate at Cromatic
Founding Staff Engineer at Axia Medicine
Associate Director/Director, Digital Innovation Alchemist at Xilio Therapeutics
Founding Technical Research Roles at Prima Mente
🙋 🙏 Community Asks
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.
🙏 Thank you for being a BiB Weekly reader!
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!
Vince Alessi
BiB Editor in Chief
 | A guest post by
|