This article provides a systematic evaluation of four advanced, commercially available high-throughput spatial transcriptomics (ST) platforms with subcellular resolution: Stereo-seq v1.3, Visium HD FFPE, CosMx 6K, and Xenium 5K. To establish a robust benchmark, the researchers generated serial tissue sections from human cancer samples (colon adenocarcinoma, hepatocellular carcinoma, and ovarian cancer) and used CODEX protein profiling and single-cell RNA sequencing (scRNA-seq) as orthogonal ground truth datasets. The study comprehensively assesses platform performance across technical metrics such as sensitivity, spatial accuracy, and diffusion control, as well as downstream analytical capabilities including cell segmentation, cell type annotation, and spatial clustering. The resulting multi-omics dataset is publicly accessible via the SPATCH web server, offering a valuable resource for future computational and biological discoveries in the spatial omics field.

References:

• Ren P, Zhang R, Wang Y, et al. Systematic benchmarking of high-throughput subcellular spatial transcriptomics platforms across human tumors[J]. Nature Communications, 2025, 16(1): 9232.

The paper describes the development and application of a computational pipeline for spatial mechano-transcriptomics, which is a new framework for jointly analyzing transcriptional and mechanical signals at single-cell resolution within tissues. This framework integrates spatial transcriptomics data with an image-based mechanical force inference approach, such as the Variational Method of Stress Inference (VMSI), to quantify cellular properties like pressure and tension. Applied to the developing mouse embryo, the analysis demonstrates that tissue compartment boundaries are characterized by elevated cell–cell junctional tension, a finding robustly supported by biophysical simulations. Furthermore, the study utilizes ligand-receptor analysis and structural equation modeling (gSEM) to identify gene modules and signaling pathways, including ephrin signaling, that are associated with and potentially regulate these mechanical properties, often exhibiting nonlinear expression patterns in response to mechanical changes.

References:

• Hallou A, He R, Simons B D, et al. A computational pipeline for spatial mechano-transcriptomics[J]. Nature methods, 2025: 1-14.

The paper introduces Flexynesis, a novel deep learning toolkit designed for the bulk integration of multi-omics data, primarily aimed at precision oncology applications. The authors developed Flexynesis to address the limitations of existing deep learning methods, which often lack transparency, modularity, and easy deployability for clinical and preclinical research. Flexynesis offers a comprehensive framework that streamlines data processing, allows users to choose from various deep learning or classical machine learning architectures, and supports diverse tasks like regression, classification, survival modeling, and biomarker discovery. The article showcases the tool's versatility through several use cases, including single-task and multi-task learning, and demonstrates its accessibility across platforms like PyPi, Bioconda, and the Galaxy Server. Ultimately, Flexynesis is presented as a flexible, reproducible, and accessible solution to enhance the utility of deep learning in multi-omics data analysis.

References:

• Uyar B, Savchyn T, Naghsh Nilchi A, et al. Flexynesis: A deep learning toolkit for bulk multi-omics data integration for precision oncology and beyond[J]. Nature Communications, 2025, 16(1): 8261.

The paper details a study investigating the distinct ways glioblastoma (GBM) tumors metabolize glucose compared to healthy brain tissue. Using stable isotope tracing in human patients and mice, the researchers determined that while the cortex uses glucose primarily for physiological functions like the TCA cycle and neurotransmitter synthesis, aggressive GBM redirects glucose-derived carbons to produce nucleotides and NAD/NADH to support proliferation and invasion. Crucially, the study identifies that many gliomas prioritize environmental serine uptake over glucose-driven synthesis, a metabolic vulnerability that can be exploited, as restricting dietary serine/glycine was shown to slow tumor growth and enhance the efficacy of standard chemoradiation treatment. These findings uncover a key metabolic difference in brain cancer that suggests novel strategies for targeted therapy, potentially paving the way for precision dietary interventions to improve GBM outcomes.

References:

• Scott A J, Mittal A, Meghdadi B, et al. Rewiring of cortical glucose metabolism fuels human brain cancer growth[J]. Nature, 2025: 1-10.

The paper outlines the creation of a global, multiscale map of human subcellular architecture in U2OS osteosarcoma cells, integrating two complementary data sources: biophysical protein interactions determined by affinity purification–mass spectrometry (AP–MS) and protein localization images from immunofluorescence (IF). The researchers employed a self-supervised machine learning model to fuse these modalities, resolving 275 distinct molecular assemblies spanning a broad range of physical sizes, which they systematically validated using whole-cell size-exclusion chromatography (SEC–MS). This reference map was annotated with the help of large language models (LLMs) and applied to diverse biological problems, including 3D structural modeling, assigning unexpected functions to hundreds of proteins, analyzing cell-type specificity and multi-localization, and interpreting paediatric cancer mutations that converge on specific protein assemblies. The study culminates in a reference platform and open-source toolkit for structural and functional cell biology.

References:

• Schaffer L V, Hu M, Qian G, et al. Multimodal cell maps as a foundation for structural and functional genomics[J]. Nature, 2025: 1-10.

The paper introduces CellNEST, a novel computational model utilizing Graph Attention Networks (GAT) and Deep Graph Infomax (DGI) to overcome challenges in detecting cell-cell communication (CCC) from spatial transcriptomics data, which is essential for understanding diseases like cancer. Existing methods often suffer from high false-positive rates and only detect single ligand-receptor pairs, while CellNEST accurately identifies both single CCC events and complex relay networks of communication by leveraging spatial context and pattern recognition. The paper demonstrates CellNEST's superior performance across various cancer and healthy tissues, including its ability to localize specific signals, like T cell homing in lymph nodes or aggressive communication in pancreatic ductal adenocarcinoma (PDAC), even predicting subtype-specific CCC in PDAC patients. Furthermore, the authors offer CellNEST-Interactive, a web-based tool for visualizing these complex spatial communication networks.

References:

• Zohora F T, Paliwal D, Flores-Figueroa E, et al. CellNEST reveals cell–cell relay networks using attention mechanisms on spatial transcriptomics[J]. Nature Methods, 2025: 1-15.

The article details an experiment demonstrating that the brain anticipates virtual infection threats, triggering a measurable immune response before actual physical contact with a pathogen occurs. Researchers used virtual reality (VR) to expose participants to avatars displaying signs of infection, measuring neural activity through EEG and fMRI, and monitoring changes in innate lymphoid cells (ILCs). The results show that the peripersonal space (PPS) system and the salience network detect these potential threats, leading to ILC activation similar to that caused by a flu vaccine. This neuro-immune cross-talk is connected to the hypothalamic–pituitary–adrenal (HPA) axis and is mediated by a complex, nonlinear relationship involving hormones and other signaling factors, suggesting an integrated, proactive defense mechanism in humans against

References:

• Trabanelli S, Akselrod M, Fellrath J, et al. Neural anticipation of virtual infection triggers an immune response[J]. Nature neuroscience, 2025: 1-10.

The article investigates the role of adipose tissue macrophages (ATMs) in the context of aging and chronic inflammation, referred to as "inflammaging." Using single-cell RNA sequencing and mouse models, the researchers identify diverse ATM subsets, including a novel population called nerve-associated macrophages (NAMs), which are characterized by the marker CD169 and decrease with age, particularly in female mice. The study demonstrates that the depletion of these CD169+ NAMs impairs the metabolic function of adipose tissue by disrupting lipolysis and promoting catecholamine resistance, ultimately exacerbating age-related inflammation. Furthermore, the research defines a unique inflammatory subset, age-associated macrophages (AAMs), which accumulates significantly in aged visceral adipose tissue, linking macrophage remodeling to age-related tissue dysfunction.

References:

• Gonzalez-Hurtado E, Leveau C, Li K, et al. Nerve-associated macrophages control adipose homeostasis across lifespan and restrain age-related inflammation[J]. Nature Aging, 2025: 1-16.

The paper details research on treating highly lethal childhood cancer neuroblastoma. The study investigates combining the existing drug difluoromethylornithine (DFMO), which inhibits polyamine synthesis, with a proline- and arginine-free diet to enhance therapeutic outcomes. Researchers found that this combined approach drastically depleted the polyamine precursor ornithine, leading to a significant survival benefit and tumor differentiation in mouse models. Mechanistically, this depletion causes specific ribosome stalling at codons ending in adenosine, preferentially suppressing the translation of cell cycle genes and promoting the translation of differentiation-associated genes, thus reprogramming the tumor proteome. Ultimately, the research proposes that targeting codon usage via metabolic stress is a viable strategy to induce therapeutic differentiation in pediatric cancers.

References:

• Cherkaoui S, Turn C S, Yuan Y, et al. Reprogramming neuroblastoma by diet-enhanced polyamine depletion[J]. Nature, 2025: 1-9.

This review addresses the major challenge in cancer immunotherapy posed by "cold" tumors, which resist treatment due to low immune cell infiltration and an immunosuppressive tumor microenvironment (TME). The authors meticulously examine the molecular and cellular mechanisms of this resistance, categorizing them into three "Cs": camouflage (impeding immune priming and infiltration), coercion (suppressing immune function), and cytoprotection (resisting inflammatory death). Crucially, the text evaluates numerous therapeutic strategies aimed at converting these refractory "cold" tumors into immunologically "hot" entities, discussing advancements such as epigenetic reprogramming, metabolic interventions, and the innovative use of biomaterials for localized drug delivery. Ultimately, the review frames the transformation of cold tumors to hot tumors as a critical pathway to broaden the efficacy of immune checkpoint inhibitors and other advanced immunotherapies.

References:

• Liu Y T, Wang Y L, Wang S, et al. Turning cold tumors into hot tumors to ignite immunotherapy[J]. Molecular Cancer, 2025, 24(1): 254.