Hony Ventures backs AI-driven biotech startups in Asia and China

#GRAL has been red hot, and I think there’s more room to go because this company is leading the Biotech AI Revolution, which presently feels like the most undervalued and underappreciated vertical of the AI Boom. Here’s the bull thesis in a nutshell... GRAIL is a biotech / diagnostics company focused on multi-cancer early detection (MCED) using liquid biopsy / methylation / cell-free DNA signal technology. Its flagship product is Galleri -- a blood test intended to detect signals of many cancers (beyond the “usual suspects”) in asymptomatic individuals. Basically, an early cancer screen. Big potential. But the growth unlock here is in AI. Galleri is basically a pattern-recognition problem at genomic scale -- and AI is the key to cracking it. By applying machine learning to massive cfDNA methylation datasets, GRAIL can detect faint cancer signals others can’t. Every new test feeds more data into the system, creating a flywheel where more data -> better AI models -> higher accuracy -> wider adoption -> even more data. This makes GRAIL not just a biotech company, but an AI-powered diagnostics platform with the potential to become the standard infrastructure for early cancer detection. If it succeeds, it could own one of the most valuable AI datasets in healthcare, capture payer adoption, and re-rate from speculative biotech to AI-healthcare leader -- a much higher multiple story. AI turns GRAIL’s cancer test from a moonshot into a scalable, defensible, and potentially transformative healthcare business. GRAL appears to be in the early innings of that transformation now. As it continues, the stock could keep blasting higher, in my view.

What if AI could help us not just treat disease but reverse it at the cellular level? Harvard Medical School has just introduced PDGrapher, a breakthrough AI model published in Nature Biomedical Engineering. Unlike older methods that ask “what does this drug do?”, PDGrapher asks a better question: “What combination of targets will restore a diseased cell to health?” Why this matters for responsible AI in healthcare Most drug discovery is slow, costly, and focused on one target at a time. PDGrapher does three things differently: Direct predictions: It identifies therapeutic targets and combinations that shift diseased cells back to healthy ones. Scalable & efficient: Trains 25 times faster than other AI methods, reducing waste and cost. Clinically grounded: Validated across 11 cancers and found both known targets (like VEGFR2 in lung cancer) and overlooked ones (like TOP2A), opening new doors for existing drugs. And here’s the responsible part: Harvard has made PDGrapher open-source, so every researcher not just big pharma can build on it. The implications More personalized therapies for patients Faster repurposing of existing drugs New strategies for “undruggable” diseases A model for how AI can be used responsibly: transparent, collaborative, clinically relevant

The "Undruggable" Myth Just Collapsed. Here’s What I’ve Learned⚡ Two years ago, I was virtually screening 200,000+ compounds against Aurora Kinase A-N-myc one of cancer’s iconic “undruggable” targets. Today, the landscape is transformed and it’s not because of static structure prediction. Proteins are not static in nature they’re living, dynamic molecules, flexing through countless conformations. The truly cryptic druggable sites those elusive allosteric pockets appear only as proteins move. WHERE THE BREAKTHROUGH HAPPENED: → AI-augmented Molecular Dynamics (MD) Simulations: We can now map thousands of conformations and observe binding pockets in action, detecting transient, druggable sites invisible to any single static structure. → Machine Learning for Pocket Detection: Advanced ML models rapidly identify and rank these cryptic sites across vast MD datasets, amplifying what humans and classical tools would miss. → Allosteric Modulation: We now have the ability to model and drug targets like MYC, KRAS, and tau once labeled “undruggable” by leveraging protein dynamism. PLATFORMS MAKING WAVES: 🚀 Atomwise - scaling conformational ensemble docking and cryptic pocket discovery 🚀 Gain Therapeutics - AI+MD pipelines finding allosteric sites in neurological targets 🚀 Optibrium - transitioning towards AI-driven cryptic site prediction with dynamic input 🚀 Community-wide leveraging resources like GPCRmd, AlphaFold DB (for starting models), and long-timescale MD platforms Working on the AURKA-N-myc interface taught me: “undruggable” proteins aren’t impossible they’re just misunderstood. By running long-timescale MD (over 1.1 μs), I captured conformational changes that revealed potent leads each disrupting the oncogenic axis. Static snapshots didn’t cut it: dynamics changed the game. Drug discovery isn’t about finding more targets it’s about realizing that targetability itself is a function of protein motion and flexibility. → Static analysis → Dynamic, ensemble-based exploration → Orthosteric focus → Allosteric and cryptic modulation → One-size-fits-all → Precision, context-aware therapeutics Within years, every pharma pipeline will make AI-enhanced conformational dynamics a standard tool. Companies that embrace this shift will lead the next decade of innovation. With unlimited compute resources, which “undruggable” protein would you explore for cryptic, dynamic pockets? Let’s challenge traditional dogma and redefine what’s “druggable.” #AIinDrugDiscovery #Biotechnology #ProteinDynamics #Innovation #DrugDevelopment #Bioinformatics #PersonalizedMedicine #Pharma #HealthTech #ComputationalBiology

⛑️ Thaumatec HealthTech Industry Update⛑️ 😎 This week – Applicable HealthTech in BioTech 😎 Welcome to our update, where we share the latest news and updates and interesting insights from the HealthTech Industry. Health Technology topics e.g. from Medical Devices and Digital Health and as well IoMT Solutions which is one bracket here are applicable in the biotech area and industry too and included in advanced diagnostics, personalized medicine, AI and machine learning, gene editing and CRISPR diagnostics, regenerative medicine, genetic diagnostics, genome sequencing, and digital health tools such as wearable devices for patient monitoring. Here some important connections between Biotech and HealthTech: 🩸 Biotech and Artificial Intelligence (AI) and Machine Learning 🩸 Biotech Data and Integration 🩸 Biotech and Medical Devices 🩸 Biotech and Laboratories These technologies drive innovation in biotech by enabling earlier disease detection, customizing treatments to individual genetic profiles, accelerating drug discovery, and improving patient outcomes through real-time data and remote care solutions. #HealthTech #biotech #Healthcare #iomt #medicaldevices #digitlhealth Interested? Have a look on some details in the Article of our THAUMATEC TECH GROUP knowledge Database: https://lnkd.in/ggqNhatn

At Servier Symphogen, we are combining the biological strength of antibodies with the power of artificial intelligence (AI) to shape the future of cancer research. AI is becoming a true research partner, helping us to: ·                Identify new targets in cancer cells ·                Design promising molecules ·                Predict human doses earlier ·                Accelerate clinical trial processes Alongside advanced antibody formats such as ADCs and multispecifics, this approach enables us to innovate faster and with greater precision – ultimately aiming to bring new treatments to patients sooner. 👉 Read the full article in Børsen: https://lnkd.in/d4mT8_FK #Oncology #AI #Antibodies #ADCs #Innovation #LifeScience #Servier#Research #Collaboration

🚨🇺🇸🧬 HARVARD AI MODEL IDENTIFIES GENES AND DRUG COMBINATIONS TO REVERSE DISEASE IN CELLS Techno-skeptic narrative: AI models in drug discovery are fundamentally flawed black boxes that researchers shouldn't trust. These algorithms understand nothing about chemistry and make purely statistical correlations without a scientific basis. Most AI models excel only on retrospective benchmarks and rarely deliver actual prospective value in discovering real drug leads. As AI models often overfit datasets and struggle in real human systems, claims of restoring health in diseased cells may be premature without rigorous wet‑lab and clinical tests. Techno-optimist narrative: Harvard's PDGrapher AI model is a breakthrough tool that actually reverses disease in cells by precisely targeting multiple pathways simultaneously. Unlike traditional single-target approaches, this advanced tool identifies the optimal gene combinations to restore healthy cell function. It has already validated known cancer targets and discovered new, previously overlooked ones with 35% greater accuracy than competing models. It offers a strong promise to significantly accelerate drug discovery — especially for complex diseases like cancer.

𝗔 𝗖𝗼𝗻𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗼𝗻 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗖𝗘𝗢: 𝗔𝗱𝘃𝗮𝗻𝗰𝗶𝗻𝗴 𝘁𝗵𝗲 𝗙𝘂𝘁𝘂𝗿𝗲 𝗼𝗳 𝗖𝗹𝗶𝗻𝗶𝗰𝗮𝗹 𝗧𝗿𝗶𝗮𝗹𝘀 |   LabConnect CEO Wes Wheeler recently sat down with the team at Tomorrow’s World Today to discuss how a lab-agnostic model, purpose-built software, and disciplined process design are transforming global clinical research.    In the podcast, Wes breaks down how LabConnect enables 𝘀𝗽𝗲𝗲𝗱, 𝘀𝗰𝗮𝗹𝗲, 𝗮𝗻𝗱 𝗳𝗹𝗲𝘅𝗶𝗯𝗶𝗹𝗶𝘁𝘆 across seven key regions with a tech-enabled ecosystem, converting specialty testing into submission-ready data.   Looking ahead, Wes explores how #AI, 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴, 𝗮𝗻𝗱 𝗱𝗶𝗴𝗶𝘁𝗮𝗹 𝘁𝘄𝗶𝗻𝘀 are reshaping clinical development—from modeling preclinical safety to identifying recruitable patient populations and flagging anomalies before they impact results. With cell and gene therapies, antibody-drug conjugates, and #radiopharma driving the next wave of precision medicine, he shares how LabConnect’s innovation mindset keeps trials moving forward.   🎧 𝗟𝗶𝘀𝘁𝗲𝗻 𝘁𝗼 𝘁𝗵𝗲 𝗳𝘂𝗹𝗹 𝗲𝗽𝗶𝘀𝗼𝗱𝗲: Spotify: https://lnkd.in/gzJaxkRu Online: https://lnkd.in/g8iW2Tfj Your favorite podcast app: https://lnkd.in/gK3pnVpm

Scalable Antibody Affinity Maturation Prediction via Multi-Scale Graph Neural Networks The proposed framework leverages multi-scale graph neural networks (MGNNs) to predict antibody affinity maturation trajectories, offering a 10x improvement over existing computational methods. This technology promises to accelerate antibody drug discovery, potentially impacting the $200 billion biopharmaceutical market and revolutionizing personalized immunotherapy. The system integrates sequence, structural, and interaction data within a unified graph representation, enabling accurate prediction of affinity changes during iterative mutagenesis. We implement MGNNs capturing hierarchical antibody features—amino acid residues, complementarity determining regions (CDRs), and full antibody structures. Training involves a novel loss function integrating structure prediction accuracy, binding affinity RMSD, and evolutionary conservation metrics, utilizing extensive antibody sequence datasets alongside experimental binding data from the MabSelect SuRe platform. Validation is performed using a https://lnkd.in/grVhJgmn

In today's Where Tech Meets Bio: 🤝 BenchSci and Thermo Fisher Scientific partner to build AI tools for experimental design, reagent selection, and literature analysis on ASCEND. 🧬 Arc Institute and collaborators report programmable bridge recombinases editing up to ~1 Mb of human DNA in Science. 📚 Proscia adds PictorLabs Inc’ virtual staining for digital H&E, IHC, and special stains from unstained slides. 🧬 Baker's Lab unveils RFdiffusion3, a unified protein-design diffusion model achieving atom-level accuracy with lower compute. 💰 Merck expands its Variational AI partnership to up to $349M to pursue two challenging small-molecule targets. 🤝 NetraMark teams with a U.S. medical center to stratify glioblastoma via CSF proteomics and explainable AI. 📒 GSK taps Pi Health to run a global Phase 2 oncology trial on its AI-driven platform. 💰 Manas AI raises a $26M seed extension; co-founders Siddhartha Mukherjee and Reid Hoffman name Ujjwal Singh CTO. 🚀 DaltonTx emerges from stealth with £4M to build an adaptive AI platform integrated into pharma workflows. ⚙️ Brian Hie and Xiaojing Gao labs with Arc Institute release Germinal, an open-source AI framework for rapid nanobody design. 🧭 George Church and colleagues outline a practical AI roadmap for protein design in Nature Reviews Bioengineering. 🤝 Healx partners with Vuja De Sciences, adding its osteosarcoma program as HLX-4310 moves toward clinical trials. 🧠 MindWalk Corp. (formerly ImmunoPrecise Antibodies) advances an AI-designed GLP-1 longevity program combining receptor agonists with a second resilience target. 🫁 Helmholtz Munich and Parse Biosciences map lung tissue responses to 900 drugs at single-cell resolution using GigaLab. Subscribe for weekly news and deep dives: https://lnkd.in/dNaJR4Wt Garry Pairaudeau Liran Belenzon (he/him) Sydney Fenkell Joseph Caputo Joseph Geraci

What if AI could help us not just treat disease but reverse it at the cellular level? Harvard Medical School has just introduced PDGrapher, a breakthrough AI model published in Nature Biomedical Engineering. Unlike older methods that ask “what does this drug do?”, PDGrapher asks a better question: “What combination of targets will restore a diseased cell to health?” Why this matters for responsible AI in healthcare Most drug discovery is slow, costly, and focused on one target at a time. PDGrapher does three things differently: Direct predictions: It identifies therapeutic targets and combinations that shift diseased cells back to healthy ones. Scalable & efficient: Trains 25 times faster than other AI methods, reducing waste and cost. Clinically grounded: Validated across 11 cancers and found both known targets (like VEGFR2 in lung cancer) and overlooked ones (like TOP2A), opening new doors for existing drugs. And here’s the responsible part: Harvard has made PDGrapher open-source, so every researcher not just big pharma can build on it. The implications More personalized therapies for patients Faster repurposing of existing drugs New strategies for “undruggable” diseases A model for how AI can be used responsibly: transparent, collaborative, clinically relevant #Aiinhealthcare