Google DeepMind has announced the release of AlphaGenome, a new AI model designed to predict how genetic variants affect gene regulation across the entire genome. It represents a significant advancement in computational genomics by integrating long-range sequence context with base-pair resolution in a single, general-purpose architecture.
By Robert Krzaczyński
Nature, Published online: 01 July 2025; doi:10.1038/d41586-025-02016-9
The decades-old limit on how long human embryos can be grown in culture is under debate. A new road map outlines how to extend the length of culture responsibly.
Nature, Published online: 30 June 2025; doi:10.1038/d41586-025-02034-7
After a quarter of a century, the website remains an essential tool for navigating the genome and understanding its structure, function and clinical impact.
The OWASP Foundation has officially introduced the AI Testing Guide (AITG), a new open-source initiative aimed at assisting organizations in the systematic testing and security of artificial intelligence systems. This guide serves as a fundamental resource for developers, testers, risk officers, and cybersecurity professionals, promoting best practices in AI system security.
By Robert Krzaczyński
Eur J Trauma Emerg Surg. 2025 Jun 27;51(1):238. doi: 10.1007/s00068-025-02922-7.
ABSTRACT
PURPOSE: Obesity is a prevalent disease, but its influence on post-injury biology remains unclear. In this study, we aimed to characterize the independent effect of obesity on the proteomic and metabolomic signatures of trauma.
METHODS: Plasma was obtained on arrival from injured patients at a Level 1 Trauma Center and analyzed with modern mass spectrometry-based proteomics and metabolomics. Samples obtained after start of transfusion were excluded. Patients were stratified by "obesity" (body mass index [BMI]≥30 kg/m2) vs. "no obesity" (BMI < 30 kg/m2). In sub-group analyses, patients were sub-stratified by Low Injury/Low Shock (ISS < 15, base excess [BE]≥-6mEq/L) and High Injury/High Shock (ISS≥15, BE<-6). Multiple regression was used to adjust the omics data for significant covariates prior to performing ome-wide analyses.
RESULTS: There were 183 patients included (48 [26%] with obesity and 135 [74%] without). After covariate-adjustment, multiple proteins and metabolites were correlated with ISS and/or BE and were significantly different from Low Injury/Low Shock to High Injury/High Shock only in patients with obesity. This obesity-specific omics response to injury was characterized by increased inflammation, hypercoagulability, altered nitrogen metabolism, and mitochondrial dysfunction. Patients with obesity also exhibited excessive injury-provoked tissue destruction and organ damage compared to patients without obesity. In injury severity-adjusted analyses, the obesity signature consistently displayed markers of hemolysis, likely reflecting a pre-injury hemolytic propensity.
CONCLUSION: Obesity is independently associated with altered post-injury biology, which likely underlies unique pathology in trauma patients with obesity. Identifying this aberrant response to injury is the first step in developing personalized therapies for this patient population.
PMID:40576654 | DOI:10.1007/s00068-025-02922-7
Nature, Published online: 27 June 2025; doi:10.1038/d41586-025-02011-0
Several groups hope to develop artificial-intelligence models that can predict how cells behave.
Cloudflare has unveiled thirteen new Model Context Protocol (MCP) servers, enhancing the integration of AI agents with its platform. These servers allow AI clients to interact with Cloudflare's services through natural language, streamlining tasks such as debugging, data analysis, and security monitoring.
By Craig Risi
Biomolecules. 2025 May 31;15(6):803. doi: 10.3390/biom15060803.
ABSTRACT
Lung cancer is responsible for 2.21 million annual cancer cases and is the leading worldwide cause of cancer-related deaths. Specifically, lung adenocarcinoma (LUAD) is the most prevalent lung cancer subtype resulting from genetic causes; LUAD has a 15% patient survival rate due to it commonly being detected in its advanced stages. This study aimed to identify a novel biomarker signature of early-stage LUAD utilizing gene expression analysis of human lung tissue samples. Using 22 pairs of LUAD and matched normal lung microarrays, 229 differentially expressed genes were identified. These genes were networked for their protein-protein interactions, and 44 hub genes were determined from protein essentiality. Survival analysis of 478 LUAD patient samples identified four statistically significant candidates. These candidate genes' expression profiles were validated from GTEx and TCGA (347 normal, 483 LUAD samples); immunohistochemistry validated the subsequent protein presence. Through intensive bioinformatic identification and multiple validations of the four-biomarker gene signature, AGER, MGP, and PECAM1 were identified as downregulated in LUAD; SLC2A1 was identified as upregulated in LUAD. These four biologically significant genes are involved in tumorigenesis and poor LUAD prognosis, meriting their use as a clinical biomarker signature and therapeutic targets for early-stage LUAD.
PMID:40563443 | PMC:PMC12191159 | DOI:10.3390/biom15060803
JCI Insight. 2025 Jun 26:e191595. doi: 10.1172/jci.insight.191595. Online ahead of print.
ABSTRACT
Pancreatic ductal adenocarcinoma (PDAC) has a poor survival rate due to late detection. PDAC arises from precursor microscopic lesions, termed pancreatic intraepithelial neoplasia (PanIN), that develop at least a decade before overt disease--this provides an opportunity to intercept PanIN-to-PDAC progression. However, immune interception strategies require full understanding of PanIN and PDAC cellular architecture. Surgical specimens containing PanIN and PDAC lesions from a unique cohort of five treatment-naïve patients with PDAC were surveyed using spatial-omics (proteomic and transcriptomic). Findings were corroborated by spatial proteomics of PanIN and PDAC from tamoxifen-inducible KPC (tiKPC) mice. We uncovered the organization of lymphoid cells into tertiary lymphoid structures (TLSs) adjacent to PanIN lesions. These TLSs lacked CD21+CD23+ B cells compared to more mature TLSs near the PDAC border. PanINs harbored mostly CD4+ T cells with fewer Tregs and exhausted T cells than PDAC. Peri-tumoral space was enriched with naïve CD4+ and central memory T cells. These observations highlight the opportunity to modulate the immune microenvironment in PanINs before immune exclusion and immunosuppression emerge during progression into PDAC.
PMID:40569674 | DOI:10.1172/jci.insight.191595
Nature, Published online: 25 June 2025; doi:10.1038/d41586-025-01925-z
If companies do step in to support research, will they publish the results, or keep them close to their chest?
Nature, Published online: 25 June 2025; doi:10.1038/d41586-025-01929-9
Here’s where academic researchers are flocking to — and the topics they’re pursuing.
Nature, Published online: 25 June 2025; doi:10.1038/s41586-025-09202-9
Disruption of dendritic cell (DC) interstitial motility in the tumour microenvironment promotes immune evasion, and enhancement of DC interstitial motility offers a route for DC-centric immunotherapy.
Nature, Published online: 25 June 2025; doi:10.1038/d41586-025-01998-w
Tool aims to solve the mystery of non-coding sequences — but is still in its infancy.
Adv Respir Med. 2025 Jun 12;93(3):17. doi: 10.3390/arm93030017.
ABSTRACT
Circulating free DNA (cfDNA) is genetic material released from various cells into bodily fluids. Among its fractions, circulating tumor DNA (ctDNA) originates from tumor cells and reflects their genetic material, including mutations and epigenetic changes. Methods commonly employed for detecting ctDNA in blood include next-generation sequencing (NGS) and various types of PCR. The presence of ctDNA can be utilized in liquid biopsies for many diagnostic purposes related to various cancers. It is a minimally invasive method of sampling molecular compounds from tumor cells. In this paper, we focus on current knowledge regarding the liquid biopsy of blood ctDNA in the context of lung cancer, one of the leading causes of cancer-related mortality. Currently, as a clinically approved method, liquid biopsy serves as a complementary technique in NSCLC diagnostic and genetic profiling. Other applications of liquid biopsy that are still being investigated include the detection of minimal residual disease (MRD) after curative treatment and response monitoring to systemic treatment. This review discusses current and future potential directions for the development and implementation of ctDNA for patients with NSCLC.
PMID:40558116 | PMC:PMC12189613 | DOI:10.3390/arm93030017
MiniMax has introduced MiniMax-M1, a new open-weight reasoning model built to handle extended contexts and complex problem-solving with high efficiency. Built on top of the earlier MiniMax-Text-01, M1 features a hybrid Mixture-of-Experts (MoE) architecture and a novel “lightning attention” mechanism.
By Robert Krzaczyński
Trends Mol Med. 2025 Jun 23:S1471-4914(25)00118-2. doi: 10.1016/j.molmed.2025.05.007. Online ahead of print.
ABSTRACT
Gastric cancer (GC) remains a deadly disease because of late detection and limited treatment options at advanced stages. Treatment of patients with metastatic disease is based on chemotherapy, complemented by antibodies targeting HER2, VEGFR2, and more recently PD-1 or claudin 18.2. Further targets, such as FGFR2b, as well as novel drug classes including antibody-drug conjugates (ADCs) and bispecific antibodies, are promising developments in GC treatment. Despite the failure of several targeted agents, the landscape of GC therapy is evolving rapidly, facilitated by umbrella or platform precision medicine trials. The integration of next-generation sequencing and other omics techniques into molecular tumor boards, as well as functional drug testing on patient-derived models, might bring us closer to personalized oncology and ultimately improve patient survival.
PMID:40555635 | DOI:10.1016/j.molmed.2025.05.007
Front Immunol. 2025 Jun 6;16:1606125. doi: 10.3389/fimmu.2025.1606125. eCollection 2025.
ABSTRACT
BACKGROUND: Lung adenocarcinoma (LUAD) is characterized by metabolic and immune heterogeneity, driving tumor progression and therapy resistance. While G protein-coupled receptors (GPCR) signaling is known to regulate metabolism and immunity in cancers, its role in LUAD remains poorly defined. This study explores the influence of GPCR signaling on LUAD metabolism and immune landscape.
METHODS: We performed non-negative matrix factorization (NMF) clustering of GPCR signaling genes in TCGA-LUAD cohort to identify distinct molecular subgroups. A prognostic model was developed based on GPCR signaling genes using least absolute shrinkage and selection operator (LASSO) analysis and Cox regression. Differentially expressed genes were analyzed for metabolic pathway enrichment and immune infiltration. In addition, key genes within GPCR signaling were identified and validated through functional assays.
RESULTS: NMF clustering based on GPCR signaling identified three subgroups in LUAD, with cluster 3 exhibiting poorer overall survival and significant enrichment in multiple prognostic associated metabolism pathways including purine, pyrimidine, glyoxylate and dicarboxylate metabolism. Then, we developed a GPCRscore prognostic model and validated across multiple cohorts, which effectively stratified LUAD patients into distinct risk groups. High-risk LUAD patients had an immunosuppressive microenvironment and activated metabolic reprogramming. ADM was identified as a key gene in the high-risk group, correlating with tumor stage, immune suppression, and resistance to immunotherapy. Clinically, ADM was highly expressed in tumor tissues and shows elevated concentrations in the peripheral blood of patients with advanced-stage LUAD. Subsequently, we demonstrated that knock-down of ADM in LUAD cells impaired their proliferation, migration, and invasion, while also reducing the angiogenic potential of endothelial cells in vitro. Adrenomedullin promoted LUAD progression in a murine metastasis model. Further, adrenomedullin inhibited CD8+ T cells proliferation, induced exhaustion, and impaired cytotoxic function. Finally, drug sensitivity and cell viability analysis showed LUAD patients with high levels of ADM exhibited sensitivity to the treatment of Staurosporine and Dasatinib.
CONCLUSIONS: In summary, this study reveals the pivotal role of GPCR signaling particularly mediated by ADM in orchestrating metabolic reprogramming and immune modulation in LUAD. ADM emerges as a potential predictive biomarker and therapeutic target, offering valuable implications for optimizing strategies.
PMID:40547013 | PMC:PMC12179119 | DOI:10.3389/fimmu.2025.1606125
When testing an AI model, it’s hard to tell if it is reasoning or just regurgitating answers from its training data. Xbench, a new benchmark developed by the Chinese venture capital firm HSG, or HongShan Capital Group, might help to sidestep that issue. That’s thanks to the way it evaluates models not only on the ability to pass arbitrary tests, like most other benchmarks, but also on the ability to execute real-world tasks, which is more unusual. It will be updated on a regular basis to try to keep it evergreen.
This week the company is making part of its question set open-source and letting anyone use for free. The team has also released a leaderboard comparing how mainstream AI models stack up when tested on Xbench. (ChatGPT o3 ranked first across all categories, though ByteDance’s Doubao, Gemini 2.5 Pro, and Grok all still did pretty well, as did Claude Sonnet.)
Development of the benchmark at HongShan began in 2022, following ChatGPT’s breakout success, as an internal tool for assessing which models are worth investing in. Since then, led by partner Gong Yuan, the team has steadily expanded the system, bringing in outside researchers and professionals to help refine it. As the project grew more sophisticated, they decided to release it to the public.
Xbench approached the problem with two different systems. One is similar to traditional benchmarking: an academic test that gauges a model’s aptitude on various subjects. The other is more like a technical interview round for a job, assessing how much real-world economic value a model might deliver.
Xbench’s methods for assessing raw intelligence currently include two components: Xbench-ScienceQA and Xbench-DeepResearch. ScienceQA isn’t a radical departure from existing postgraduate-level STEM benchmarks like GPQA and SuperGPQA. It includes questions spanning fields from biochemistry to orbital mechanics, drafted by graduate students and double-checked by professors. Scoring rewards not only the right answer but also the reasoning chain that leads to it.
DeepResearch, by contrast, focuses on a model’s ability to navigate the Chinese-language web. Ten subject-matter experts created 100 questions in music, history, finance, and literature—questions that can’t just be googled but require significant research to answer. Scoring favors breadth of sources, factual consistency, and a model’s willingness to admit when there isn’t enough data. A question in the publicized collection is “How many Chinese cities in the three northwestern provinces border a foreign country?” (It’s 12, and only 33% of models tested got it right, if you are wondering.)
On the company’s website, the researchers said they want to add more dimensions to the test—for example, aspects like how creative a model is in its problem solving, how collaborative it is when working with other models, and how reliable it is.
The team has committed to updating the test questions once a quarter and to maintain a half-public, half-private data set.
To assess models’ real-world readiness, the team worked with experts to develop tasks modeled on actual workflows, initially in recruitment and marketing. For example, one task asks a model to source five qualified battery engineer candidates and justify each pick. Another asks it to match advertisers with appropriate short-video creators from a pool of over 800 influencers.
The website also teases upcoming categories, including finance, legal, accounting, and design. The question sets for these categories have not yet been open-sourced.
ChatGPT-o3 again ranks first in both of the current professional categories. For recruiting, Perplexity Search and Claude 3.5 Sonnet take second and third place, respectively. For marketing, Claude, Grok, and Gemini all perform well.
“It is really difficult for benchmarks to include things that are so hard to quantify,” says Zihan Zheng, the lead researcher on a new benchmark called LiveCodeBench Pro and a student at NYU. “But Xbench represents a promising start.”
Login