#93 - Rosalind returns, repurposed rejects & rebel replication
The coffee break biotech roundup, by SomX.
Hello my fellow mavericks,
This week: Boehringer breaks ranks and builds its AI capabilities rather than buying them, OpenAI debuts a drug discovery model named after the woman who never got the credit she deserved, a Finnish startup defies the odds and pips Harvard and Stanford to a quantum biology prize, Formation Bio tears up the industry playbook to beat the real clinical bottleneck, and a bacterial enzyme goes rogue and writes DNA from its own amino acids.
Vive la révolution,
Dodo
If there’s anything you’d like to see in future editions of Biotech Dodo, send us a message.
Discover 🔍
💂 Boehringer Ingelheim launches AI Accelerator in London (Boehringer Ingelheim): The new King’s Cross hub is the German pharma giant’s fourth computational AI centre and joins a roster of sites in Austria, Germany, and the US under a £150M 10-year investment. Plonked in the Knowledge Quarter alongside the Francis Crick Institute and the Alan Turing Institute, it will focus on patient journey mapping, biological mechanism discovery, and disease driver identification, with a goal of 50 AI experts in place by end of 2027.
Our take: Most pharma is busily snapping up AI capabilities – Novo Nordisk just partnered with OpenAI, Sanofi acquired Exscientia for $1.2B, and Lilly bet $2.75B on Insilico – but Boehringer is having a crack at the rather more patient business of building foundational capability from scratch. Fifty experts by 2027 and £15M a year is a modest opening wager but, if it pays off, the intellectual property and institutional knowledge stay firmly in-house.
🩻 OpenAI debuts AI model GPT-Rosalind to speed up drug discovery (TechTarget): Named after Rosalind Franklin – whose X-ray crystallography helped crack the structure of DNA – the new domain-specific model combines multi-step reasoning across chemistry, genomics, and protein engineering with access to more than 50 scientific tools and databases. Amgen, Moderna, and Thermo Fisher are among the early partners, though access is firmly gated behind a vetted trusted access programme.
Our take: OpenAI’s life sciences research lead Joy Jiao has been rather candid that GPT-Rosalind cannot invent new drugs on its own; it is a reasoning layer for hypothesis generation, literature synthesis, and experimental planning – nothing more. The limited access reflects genuine biosecurity concerns, but also (rather conveniently) positions the model as premium and exclusive at a moment when OpenAI is hungry for enterprise revenue. A fitting name – perhaps this time she'll get the credit she is owed.
⚛️ Helsinki’s Algorithmiq wins €1.7 million prize for quantum-enabled light-sensitive cancer drug discovery (EU-Startups): The sole winner of Wellcome Leap’s 2.5-year Quantum for Bio challenge, pipping finalists from Harvard, Stanford, and Oxford to the post, demonstrated a quantum-classical framework for simulating a photosensitiser drug currently in Phase 2 trials. The framework ran on up to 100 qubits on existing IBM hardware and was validated against state-of-the-art classical methods.
Our take: Quantum advantage in drug discovery has been a theoretical promise for the better part of two decades, consistently done in by hardware that couldn’t deliver it. Algorithmiq’s winning framework had to meet three simultaneous requirements: executable on current hardware, biologically meaningful, and validated against classical methods under realistic constraints. That last condition is the one the field has been dodging – passing it, on machines that actually exist, is no small thing.
🎯 This Sam Altman-backed $1.8 billion startup bets AI can get drugs through clinical trials faster (Forbes): Formation Bio has raised $615M to snap up stalled drug candidates and frogmarch them through clinical trials with AI. The New York startup targets pre-Phase 2 drugs, where only 30% succeed, arguing that clinical development rather than discovery is the industry’s real bottleneck. The company has already licensed one asset to Sanofi for $630M, with four more in trials. Backers include Andreessen Horowitz, Sequoia, and OpenAI’s Sam Altman.
Our take: Formation’s osteoarthritis drug had already been through three trials across 800 patients, without clear clinical evidence of impact, and was quietly gathering dust. AI didn’t discover anything new – it found signals in data that already existed, establishing that increased cartilage growth from the drug did in fact translate into a lower five-year risk of needing a knee replacement. That’s a rather different proposition from drug discovery AI: not generating new therapies, but rescuing value from ones that were close to being written off. If Formation can do that reliably at scale, the pipeline of abandoned post-Phase 1 assets starts to look rather less like a graveyard.
And finally…
🧬 Scientists stunned by ‘fundamentally new way’ life produces DNA (Science): Stanford researchers have discovered DRT3, a bacterial defence system harbouring an enzyme, Drt3b, that synthesises DNA without a nucleic acid template – instead, using its own amino acids as a guide. Widespread among bacteria, DRT3 appears to protect against viral infection, though precisely how it does so remains a mystery. The researchers hope it can be engineered into a tool for custom DNA synthesis.
Our take: Biology’s “central dogma” – that information flows from nucleic acids to protein, never the other way around – has held for decades. Drt3b doesn’t rewrite it exactly, but a protein serving as its own DNA template has no precedent in biology, and the researchers are candid that they don’t yet know how DRT3 actually defeats viruses. The CRISPR parallel is already being drawn. It, too, started as an obscure bacterial defence mechanism, with applications nobody had imagined.
Tune in 🎧
🎰 The medications that can trigger sex and gambling addictions: A BBC investigation into dopamine agonists examines how roughly 1 in 6 patients develop impulse control disorders, and why many were never warned of the risk.
🐺 Ancient Egyptian DNA, Cheddar Man and wolves as pets: Brian Cox and two ancient DNA specialists field listener questions on how far back genomic analysis can reach, and what early human genetic diversity tells us about survival.
☄️ The Hitchhiking Microbe’s Guide to the Galaxy: Researchers test the plausibility of lithopanspermia – the hypothesis that microbes hitchhike between planets on meteorites – by firing them from a gun to simulate the forces involved.
Apply ✍️
🛠️ Development Engineer, Bespak: Want to put the device in drug delivery? You’ll support product design from concept to commercialisation, run CAD simulations, conduct NGI testing, and steer precision-moulded parts through performance and regulatory requirements.
💯 (Fixed Term) Analyst, Quality Control, Chemistry & Stability, Moderna: Can you keep mRNA vaccines in good shape? You’ll run HPLC, UPLC, and stability testing across the product lifecycle, manage LIMS workflows, and keep the brand-new Harwell facility audit-ready.
🧪 Research Scientist, Arthur Edward: Think you can coax microbes into industrial submission? You’ll develop and optimise cell culture processes from lab to pilot scale, apply strain engineering techniques, and drive tech transfer through to commercial manufacture.
RSVP 📆
🏥 28.04 | The Future of Prevention: Who Pays, Who Delivers, Who Benefits? | London, UK: A closed-door forum bringing together senior NHS, government, healthtech, and employer leaders to tackle the UK’s prevention agenda – from funding models to delivery at scale. Hosted by SomX (that’s us!) and Neko Health.
🇰🇷 29.04 | Bio Korea 2026 | Seoul, South Korea: One of Asia’s leading life sciences gatherings, bringing together biopharma, biotech, and healthtech leaders for partnering, investment, and industry dialogue.
👮13.05 | Cybersecurity and clinical risk management in the era of AI | Royal Society of Medicine, London, UK: A half-day event for clinicians, healthcare leaders, and digital health professionals navigating AI's security and clinical risks with practical guidance from NHS CCIOs on deploying it safely and compliantly.
Got news, jobs or events you think are worth coo-ing over? Post an event here, or email us at biotechdodo@substack.com!
We’re SomX – a communications and creative agency trusted by biotech, pharma and healthtech pioneers. We craft strategy, content, PR and design that translate complex biology into compelling stories and investor‑ready excitement. Get in touch to amplify your science beyond the lab bench.