Recent Breakthroughs in Research and Development

Big breakthrough: A few months my lab at MIT introduced SPARKS, our autonomous scientific discovery model. Since then we have demonstrated applicability to broad problem spaces across domains from proteins, bio-inspired materials to inorganic materials. SPARKS learns by doing, thinks by critiquing itself & creates knowledge through recursive interaction; not just with data, but with the physical & logical consequences of its own ideas. It closes the entire scientific loop - hypothesis generation, data retrieval, coding, simulation, critique, refinement, & detailed manuscript drafting - without prompts, manual tuning, or human oversight. SPARKS is fundamentally different from frontier models. While models like o3-pro and o3 deep research can produce summaries, they stop short of full discovery. SPARKS conducts the entire scientific process autonomously, generating & validating falsifiable hypotheses, interpreting results & refining its approach until a reproducible, fully validated evidence-based discovery emerges. This is the first time we've seen AI discover new science. SPARKS is orders of magnitude more capable than frontier models & even when comparing just the writing, SPARKS still outperforms: in our benchmark evaluation, it scored 1.6× higher than o3-pro and over 2.5× higher than o3 deep research - not because it writes more, but because it writes with purpose, grounded in original, validated compositional reasoning from start to finish. We benchmarked SPARKS on several case studies, where it uncovered two previously unknown protein design rules: 1⃣ Length-dependent mechanical crossover β-sheet-rich peptides outperform α-helices—but only once chains exceed ~80 amino acids. Below that, helices dominate. No prior systematic study had exposed this crossover, leaving protein designers without a quantitative rule for sizing sheet-rich materials. This discovery resolves a long-standing ambiguity in molecular design and provides a principle to guide the structural tuning of biomaterials and protein-based nanodevices based on mechanical strength. 2⃣ A stability “frustration zone” At intermediate lengths (~50- 70 residues) with balanced α/β content, peptide stability becomes highly variable. Sparks mapped this volatile region and explained its cause: competing folding nuclei and exposed edge strands that destabilize structure. This insight pinpoints a failure regime in protein design where instability arises not from randomness, but from well-defined physical constraints, giving designers new levers to avoid brittle configurations or engineer around them. This gives engineers and biologists a roadmap for avoiding stability traps in de novo design - especially when exploring hybrid motifs. Stay tuned for more updates & examples, papers and more details.

The last two days have seen two extremely interesting breakthroughs announced in quantum computing. There is a long path ahead, but these both point to the potential for dramatically upscaling ambitions for what's possible in relatively short timeframes. The most prominent advance was Microsoft's announcement of Majorana 1, a chip powered by "topological qubits" using a new material. This enables hardware-protected qubits that are more stable and fault-tolerant. The chip currently contains 8 topologic qubits, but it is designed to house one million. This is many orders of dimension larger than current systems. DARPA has selected the system for its utility-scale quantum computing program. Microsoft believes they can create a fault-tolerant quantum computer prototype in years. The other breakthrough is extraordinary: quantum gate teleportation, linking two quantum processes using quantum teleportation. Instead of packing millions of qubits into a single machine—which is exceptionally challenging—this approach allows smaller quantum devices to be connected via optical fibers, working together as one system. Oxford University researchers proved that distributed quantum computing can perform powerful calculations more efficiently than classical systems. This could not only create a pathway to workable quantum computers, but also a quantum internet, enabling ultra-secure communication and advanced computational capabilities. It certainly seems that the pace of scientific progress is increasing. Some of the applications - such as in quantum computing - could have massive implications, including in turn accelerating science across domains.

Last month, a team of scientists and physicians achieved something extraordinary: they developed and delivered the first-ever personalized #CRISPR therapy to treat an infant with a life-threatening #raredisease — in just six months. A one-letter change in the baby’s DNA was corrected using a custom-built gene editor. The child, who was once facing the prospect of a liver transplant, is now steadily improving. It’s a powerful example of what’s becoming possible at the intersection of #science and #technology, urgency and purposeful ambition. And this isn’t an isolated win. Across labs, clinics, and companies, CRISPR is being used as a therapeutic modality to correct inherited disorders, engineer immune cells, disable viral DNA, and even edit entire chromosomes. New gene-editing systems—like TIGR-Tas, unveiled earlier this year—are expanding what’s possible in tissues or conditions where current tools fall short. Clinical results are emerging fast—and the pace of #innovation is only picking up. At Recursion, we’re also applying #geneediting tools like CRISPR beyond therapeutics—using the technology as a tool to better understand #biology at scale. By systematically “knocking out” thousands of individual genes and measuring how those changes affect cell behavior, we’re generating large, structured datasets that feed directly into #AI models. This is helping us uncover new biological relationships and power #drugdiscovery in ways that were previously unimaginable. What ties all of this together is a commitment to applying game-changing #innovation in service of real, urgent human needs. It signals a much-needed mindset shift in #healthcare and #biopharma: to move faster, think bigger, and tackle challenges once considered out of reach—and to truly deliver on the promise of #precisionmedicine. And we’re seeing this ambition in many other areas as well – just last week, for example, GRAIL announced more promising than ever performance stats for its #Galleri blood test for the early detection of 50+ types of #cancer. There’s still work ahead to ensure breakthroughs translate into broad, equitable impact. But this moment – this momentum – is worth pausing to recognize. We’re no longer just imagining a future where science works smarter and faster for patients. We’re building it.

Big moment for CAR-T. The “in vivo” version just cleared a major milestone. Interius BioTherapeutics dosed the first cancer patient with CAR-T cells made inside the body last fall. Now Carl June’s UPenn team has published results in Science showing how targeted nanoparticles can do the same - no external manufacturing. Why this matters: → Current CAR-T runs $500K+ per patient → Weeks to make, only at select centers → Many patients die waiting → Manufacturing failures = no treatment at all June’s method packages RNA into nanoparticles that reprogram T-cells in the patient. In monkeys - cancer cells gone in a day. In mice - tumors undetectable after 3 weeks. With Interius, EsoBiotec, and UPenn all closing in from different angles, this field isn’t “promising” anymore - it’s maturing. No cross-country blood shipments. No months-long wait. Just an infusion any cancer center could give. We’ve built Ferraris when patients needed Fords. Now the road’s finally opening.

This isn't just another incremental update. This feels like a major shift. Google's new "Willow" quantum chip just did something astounding: it solved a 150-year-old physics problem in 𝟐 𝐡𝐨𝐮𝐫𝐬. To put that in perspective, the same problem would take one of the world's fastest supercomputers years to crack. That’s a 13,000x speedup. 𝐒𝐨, 𝐰𝐡𝐚𝐭 𝐝𝐢𝐝 𝐢𝐭 𝐚𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐝𝐨? It ran a new algorithm called "Quantum Echoes" to model how atoms interact within molecules (using nuclear magnetic resonance). In simple terms, it's a brand new way of "seeing" structures in chemistry and physics that were invisible until now. 𝐖𝐡𝐲 𝐭𝐡𝐢𝐬 𝐢𝐬 𝐚 𝐛𝐢𝐠𝐠𝐞𝐫 𝐝𝐞𝐚𝐥 𝐭𝐡𝐚𝐧 𝐩𝐚𝐬𝐭 "𝐪𝐮𝐚𝐧𝐭𝐮𝐦" 𝐧𝐞𝐰𝐬: The keyword here is 𝐯𝐞𝐫𝐢𝐟𝐢𝐚𝐛𝐥𝐞. We've heard "quantum supremacy" before, but those were often abstract, one-off calculations. This is different. This is the first-ever 𝐯𝐞𝐫𝐢𝐟𝐢𝐚𝐛𝐥𝐞 𝐪𝐮𝐚𝐧𝐭𝐮𝐦 𝐚𝐝𝐯𝐚𝐧𝐭𝐚𝐠𝐞. The results can be confirmed by other quantum computers or even by real-world experiments. This is the moment quantum computing starts to move from a fascinating theory to a useful, practical tool. 𝐖𝐡𝐲 𝐬𝐡𝐨𝐮𝐥𝐝 𝐰𝐞 𝐜𝐚𝐫𝐞? This breakthrough has the potential to supercharge entire industries: >𝐃𝐫𝐮𝐠 𝐃𝐢𝐬𝐜𝐨𝐯𝐞𝐫𝐲: Simulating complex molecules could lead to new medicines designed in months, not decades. >𝐌𝐚𝐭𝐞𝐫𝐢𝐚𝐥𝐬 𝐒𝐜𝐢𝐞𝐧𝐜𝐞: We could design new materials for everything from better batteries to solar energy and nuclear fusion. >𝐀𝐫𝐭𝐢𝐟𝐢𝐜𝐢𝐚𝐥 𝐈𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞: It could give AI models tools we can't even imagine yet. If AI is the current leap forward, quantum might be the one that takes us to a completely different level, solving problems we don't even know how to define today. The age of useful quantum computing may have just begun. #QuantumComputing #Google #TechBreakthrough #AI #DrugDiscovery #MaterialsScience #Physics

A recent publication in Nature highlights an exciting breakthrough in drug delivery a needle-free insulin system that works through the skin. Researchers developed a smart, depth-responsive polymer that overcomes the long-standing challenge of transdermal insulin delivery by adapting to the skin environment and enabling efficient transport across biological barriers. The results showed effective and sustained blood glucose control in preclinical models without signs of inflammation, pointing toward a future of painless and non-invasive diabetes management. What I find particularly interesting is how this concept of stimuli-responsive polymer systems can be extended beyond insulin delivery. In areas like wound healing and urinary tract infections (UTIs), where biofilms and tissue barriers limit treatment efficacy, such smart delivery platforms could play a transformative role. Designing systems that respond to local environments (like pH or infection signals) to deliver antibiotics, nanoparticles, or even phage therapy could significantly improve targeted treatment outcomes. This kind of interdisciplinary approach combining material science with biomedical applications opens up exciting possibilities for developing next-generation therapeutic strategies. Looking forward to exploring how these concepts can be adapted to tackle real-world clinical challenges. Source: https://lnkd.in/g7JksnSX #Biotechnology #DrugDelivery #Nanomedicine #PhageTherapy #WoundHealing #UTI #Innovation

Another wild week in biotechnology: 1. Now you can track cellular gene expression patterns, using ultrasound, from outside the body. (The technology is built on gas vesicles, which I’ve written about before.) If you had said this would be a thing 10 years ago, nobody would have believed you. Mikhail Shapiro 2. Researchers made "an autobioluminescent transgenic mouse line," using bacterial DNA, that constantly emits photons. (Image below.) 3. Whole-genome sequencing of microbes floating in the air is becoming more popular. (Fun fact: Microbes travel all the way from the Sahara Desert to New York by riding motes of sand into the atmosphere.) 4. There's a new toolkit to program multicellular behaviors in Baker’s yeasts. Tom Ellis 5. Rational design of tunable, fluorescent DNA aptamers. The researchers used them to build “responsive nanostructures that light up upon activation.” 6. Scientists made sixteen new cytosine base editors, which outperform existing base editors 81.5 percent of the time. 7. A new competition is offering $30,000 in cash prizes for teams that can design the best PETase enzymes. The Align Foundation / Erika Alden DeBenedictis 8. Lab-grown sperm and eggs are likely a few years away. I’d wager this will happen within 3 years. 9. Cryo-EM structures of the measles virus polymerase mark an important step toward new therapeutics. 10. Chai-2 is “a multimodal generative model that achieves a 16% hit rate in fully de novo antibody design.” 11. A new method can eliminate multiple chrosomes inside of yeast cells, thus making them haploid. 12. A multi-adjuvant neoantigen vaccine for melanoma. 13. A single-cell multi-omics atlas of rice. Super important, considering rice is the main source of food for more than half of people on Earth. 14. SciArena ranks 23 LLMs based on their accuracy answering science questions. 15. Organoids grown in the lab now have vessels to transport blood and nutrients. 16. Researchers tracked how 214 diseases spread across Europe and Asia over the past 37,000 years by sequencing 1,313 ancient humans. 17. BioEmu is a deep learning model for emulating protein equilibrium ensembles that also incorporates “over 200 milliseconds of molecular dynamics (MD) simulations, static structures and experimental protein stabilities.” 18. A new model for IVF was “trained on…18 million time-lapse embryo images.” 19. First malaria treatment approved for infants weighing less than 4.5 kilograms.

We just crossed into real-time medicine. 7 breakthroughs this week—quiet, technical, mostly overlooked—reshape how we understand the body. There's a new foundation forming in medicine. Here’s what surfaced: 1. AI-designed drugs enter human trials DeepMind’s Isomorphic Labs is dosing patients with molecules designed entirely by artificial intelligence. This isn't molecular optimization...it's writing new biology. Incredible. 2. Quantum spin controls enzyme catalysis Proton transfer—core to metabolism—has now been shown to depend on electron spin. Life’s chemistry isn’t just probabilistic. It’s entangled. 3. Ordered water directs intracellular charge New work from the Guy Foundation reveals that structured water routes electric flow within proteins. Not passive background. Biological wiring. 4. Quantum imaging inches toward FDA approval Next-gen tools are now in preclinical testing as radiology adjuncts. We’re not just seeing clearer. We’re seeing deeper—into real-time biological state. 5. NIH opens funding for biomedical quantum sensors Ion traps. Photonic probes. The agency’s first formal push into quantum-enabled diagnostics. Measurement is no longer lagging. It’s catching up to the biology. Incredibly exciting. 6. GPT-4o leads rare disease diagnosis benchmarks Across 121 conditions, it outperforms open models in phenotypic inference. This isn’t “AI in healthcare.” This is AI thinking like a doctor—sometimes better. 7. Trial-level AI maps N-of-1 benefit curves Individual response prediction is now technically real. Predictive trajectory modeling for patients <before> they become ill. In the ER, I see patients stuck in diagnostic limbo every day. These aren’t just breakthroughs—they’re exit routes. At BellyMD, we’re building in parallel: Our first phenotype-derived therapeutic ships this week. Real-time journaling. Feedback. Signal detection. Not aspirational—operational. These 7 signals aren’t noise. They’re the new edges of medicine. Which one are you tracking?

MIT researchers made a breakthrough in quantum computing by achieving the highest-ever accuracy (99.998%) for single-qubit operations using a type of qubit called fluxonium. This is a huge step toward making reliable quantum computers. They fixed common errors by using creative techniques like circularly polarized microwave drives and "commensurate pulses," which improved speed and accuracy. These methods reduced the need for error correction, making quantum computing more efficient. Fluxonium qubits are special because they are less affected by noise, thanks to a unique design. While they were slower in the past, the new techniques allowed them to perform faster and more accurate operations. This work shows that fluxonium qubits could be a strong option for building future quantum computers. The results also help reduce errors in quantum systems, bringing practical quantum computing closer to reality. These techniques could be applied to other types of qubits too, making this breakthrough widely useful. In short, this research shows how combining physics and engineering can overcome big challenges in quantum computing, paving the way for more advanced and reliable systems.

This is massive news for Indian science,and we’re not talking about it enough. Researchers at IISc Bengaluru broke a 200 year old law of physics. They’ve discovered a “Dirac fluid” a nearly perfect liquid of electrons in ultra-clean graphene that violates the Wiedemann–Franz law by more than 200×. For nearly two centuries, every known metal followed this rule: When electrical conductivity rises, so does thermal conductivity. But in graphene, right at the Dirac point, IISc found the opposite, electrical conductivity surged while heat conductivity collapsed. At that tipping point, electrons stop behaving like individuals and flow together as a quantum liquid, revealing universal, material-independent behavior at the edge of known physics. The world has been sitting on the sidelines of quantum for too long. Now everything’s happening all at once. From room-temperature superconductors to quantum fluids in graphene, the boundaries between condensed matter, cosmology, and quantum information are collapsing into one continuous frontier. This will open pathways to design low-noise quantum circuits, heat-resilient chips, and energy-efficient transistors based on fluid-like electron motion. Dirac fluids could enable ultra-sensitive magnetometers, thermal detectors, and electrical readouts that amplify even vanishingly small signals. Establishes a universal, quantum-defined standard for conductance, a benchmark for next-gen precision instrumentation. We’re witnessing the convergence of fundamental discovery and industrial potential, and India just took a major step forward in both. Ref: Nature Physics- https://lnkd.in/gq9-2uyy