Archive
590 articles, sorted by publication time.
A UC Berkeley team has found in methane-producing archaea that the UAG codon, which usually means “stop,” is sometimes reread as a rare amino acid, opening a measurable gap in one of life’s most stable translation rules.
As artificial intelligence begins to help design proteins, viral fragments, and experimental workflows, the focus of biosecurity is not only on the models themselves, but also on a more practical bottleneck: who can order, manufacture, and leave a traceable record of synthetic DNA.
The safety of AI medical devices depends not only on whether the model is accurate, but also on whether the data has been tampered with before entering the model. An arXiv preprint proposes SAMD, a tool that attempts to use large language models to help identify attack paths that could lead to misdiagnosis or incorrect treatment.
A comparative study spanning hundreds of plants brings DNA that does not make proteins back onto the evolutionary stage: the long-preserved fragments may be ancient dials that regulate crop traits.
This Phase 1 human trial will not immediately answer whether disease can be treated. Instead, it will first ask a more fundamental question: whether immune cells worn down by aging or chronic infection can be safely metabolically reset.
A research team used generative protein design to create large single-component nanocages and confirmed their forms with cryo-EM; they remain far from clinical application, but point to a more controllable way to engineer viral shell-like structures.
Pleural mesothelioma does not grow into easily measured round masses, but spreads along the lung wall; ARTIMES, built by a Dutch team using more than 10,000 CT images, attempts to turn this long-blurry clinical judgment into more verifiable changes in volume.
The MraZ protein normally assembles like a doughnut, but it must bend and loosen part of its structure before it can attach to DNA and regulate bacterial division genes; this structural biology study brings the starting line of microbial proliferation closer to the atomic scale.
A collaboration worth up to $2.3 billion makes BTK more than just a battlefield for inhibitors; whether bexobrutideg can use “protein removal” to overcome resistance and the threshold for immune diseases will be tested in Phase 3 trials.
Legend Biotech’s LB2501 produced deep responses in a small number of patients with relapsed or refractory lymphoma, bringing renewed clinical focus to whether CAR-T can move from customized cell manufacturing to direct in vivo engineering.
Zasocitinib is not just another new immune drug; it moves AI-assisted molecular design into a clinical setting where it is compared head-to-head with an already marketed drug in the same class, putting efficacy, tolerability, and regulatory review back on the same exam paper.
A randomized deployment study of nearly 14,000 people brought medical AI from standardized tests into users’ real-time descriptions. The results offer rare field evidence, while also reminding readers that symptom interpretation remains institutionally and clinically distant from true diagnosis.