All articles tagged with #drug discovery
Generative AI can speed up biology research and drug screening, but it can also generate plausible yet false biological signals. Experts say AI outputs should be treated as testable ideas or synthetic data rather than definitive findings, and must be independently verified in real experiments. The risk varies by use, with cases like AlphaFold 3 illustrating how AI errors can distort evidence if not carefully checked.
LifeMine Therapeutics raised a $188 million Series E led by Bill Gates and Jeff Bezos (with RA Capital), reviving its fungus-based drug-discovery platform after a prior round. The platform clusters genes from about 100,000 fungal strains and will apply AI to prioritize ~1,200 targets. Its lead program LIFE-001 is a next-gen calcineurin inhibitor aimed at reducing transplant rejection toxicity; phase 1b in 12 islet-cell transplant patients and phase 2 in 150 kidney transplant patients are planned with readouts targeted for 2027. A potential $25–$30 billion market in transplantation underpins optimism, with LifeMine eyeing autoimmune indications and ongoing or future partnerships (e.g., with GSK) while rehiring furloughed staff.
Bristol Myers Squibb is expanding its Nvidia-powered infrastructure with the DGX SuperPOD to create what it calls the most powerful, energy-efficient AI platform in life sciences, enabling hybrid human-AI science to accelerate R&D and improve decision quality across its pipeline; the move extends a multi-year collaboration with Nvidia and aligns with other pharma AI partnerships in the sector.
DTU researchers combined a small printer-sized quantum computer with a generative AI model to design novel peptides that bind target proteins, producing more diverse and effective candidates—especially where data are scarce. The hybrid quantum-AI workflow could accelerate vaccine and immunotherapy development for underserved populations, but quantum hardware remains nascent and this is an early, incremental step rather than a breakthrough.
Anthropic unveiled Claude Science, an AI workbench for scientists, and signaled plans to develop its own drugs—targeting neglected diseases—while continuing to sell AI tools to biotech and pharma customers. Experts say AI can accelerate drug discovery but human oversight, data gaps, and regulatory hurdles mean any AI-designed drug is likely years away.
Researchers identify a new pathway where high dietary cholesterol activates the Ral protein, reducing liver LDL receptors and hindering clearance of LDL cholesterol. Blocking the enzyme cathepsin A (CTSA) with a small-molecule inhibitor stabilized LDL receptors and dramatically lowered blood cholesterol in mouse models, suggesting a novel, statin-independent approach and a potential CTSA-targeted treatment that could move to clinical trials, leveraging an existing CTSA inhibitor with prior human safety data.
USC researchers identified selective inhibitors of the brain inflammation enzyme cPLA2 that may lower Alzheimer’s risk, especially for APOE4 carriers; the compounds cross the blood-brain barrier, showed activity in brain cells and mouse models, and represent a promising but early-stage drug discovery path with safety and feasibility in humans to be determined.
Two AI-agent systems described in Nature—Google DeepMind's Co-Scientist and FutureHouse's Robin—use teams of AI to generate hypotheses, pick experiments, and analyze data to speed up early drug-discovery work. In initial tests they proposed plausible repurposing candidates for AML and dry AMD, with humans carrying out the experiments and validating results; while promising, most candidates require further validation in broader use.
Nature editors argue that AI agents can dramatically speed up research and drug discovery (as shown with Robin and Co-Scientist) but cannot replace humans: humans frame problems, guide experiments, validate results, and guard against AI errors, making human oversight and ethical considerations crucial as science increasingly relies on AI co-scientists.
Isomorphic Labs closed a $2.1 billion Series B, cementing its leadership in AI-driven drug discovery and prompting questions about how the funds will be deployed, pipeline progress, and the broader implications for biotech and AI investment.
The piece marks 25 years since click chemistry’s birth—reactions that join molecules with high specificity and minimal by-products—showing how this efficiency-driven approach transformed pharma, biology, and engineering and earned Sharpless a share of the 2022 Nobel Prize in Chemistry.
OpenAI launches GPT-Rosalind, a life sciences-focused AI model designed to accelerate research in genomics and biochemistry by synthesizing evidence and generating hypotheses, while keeping humans in the loop; it features enterprise-grade security and is being rolled out in a research preview to trusted customers (e.g., Amgen, Moderna, Allen Institute, Thermo Fisher) to speed health outcomes and drug discovery without replacing scientists, amid concerns about misuse.
Researchers at the Medical University of South Carolina are pursuing a treatment for familial hypercholesterolemia by reducing apoB, the backbone of LDL particles, using human iPSC-derived liver cells to screen about 130,000 compounds. A lead class lowered apoB release and cellular lipid levels in human cells and in humanized mice, bypassing the LDL receptor and potentially helping patients with defective receptors. Further work will clarify the mechanism, safety, and how such drugs could complement existing therapies like statins.
Deep underground Lechuguilla Cave hosts ancient, isolated bacteria that resist most natural antibiotics, suggesting antibiotic resistance predates humans and could inspire new drugs while offering clues on how pathogens might evolve resistance to upcoming therapies.
AI is speeding up drug discovery by screening millions of compounds to find new antibiotics against drug-resistant bacteria such as Neisseria gonorrhoeae and MRSA, and by identifying potential Parkinson's therapies through machine learning. The technology is also repurposing existing drugs for rare diseases and creating virtual disease models to test treatments, potentially slashing development time and costs—though validation, data access, and the long path to clinical use remain significant hurdles.