High-resolution spatial transcriptomics reveals that JAG1-NOTCH1 signaling drives the formation of a CAF-Epi niche during esophageal cancer progression. This niche triggers ECM remodeling and creates an immune-suppressive barrier, facilitating tumor evasion and poor prognosis.

References:

• Chang J, Lu J, Liu Q, et al. Single-cell multi-stage spatial evolutional map of esophageal carcinogenesis[J]. Cancer Cell, 2025, 43(3): 380-397. e7.

Gene expression noise, driven by transcriptional bursting and RNA degradation, creates cellular variability essential for cell fate decisions. Advanced single-cell technologies now quantify these dynamics genome-wide, revealing how noise peaks before lineage commitment in embryos.

References:

• Gene expression noise in development: genome-wide dynamics. Bax, Danique J.C. et al.Trends in Genetics, Volume 0, Issue 0

This research maps the mouse kidney's spatial and regulatory landscape across ages and sexes. It identifies sex-biased gene expression emerging during puberty, driven by hormonal pathways like PRLR/JAK/STAT. It also highlights aging-related changes and their links to human disease.

References:

• Chen S, Liu R, Mo C K, et al. Multi-omic and spatial analysis of mouse kidneys highlights sex-specific differences in gene regulation across the lifespan[J]. Nature Genetics, 2025: 1-15.

TEDD 2.0 is an expanded database for temporal gene expression during development, featuring 15 million cells across nine species. It introduces virtual gene knockout, cross-species integration, and trajectory inference to reveal cell fate decisions and conserved mechanisms.

References:

• Chan C C, Lam K K, Chen L, et al. TEDD 2.0: an advanced temporal gene expression database enabled by in-silico functional analyses for developmental mechanism investigation[J]. Science China Life Sciences, 2025: 1-15.

Researchers introduced scRank, a computational tool using single-cell transcriptomics and gene regulatory networks to identify drug-targeted cell types. They identified Ex_9 neurons as targets for SSRIs in MDD and found that Tanshinone IIA binds CTSB in macrophages.

References:

• scRank infers drug-responsive cell types from untreated scRNA-seq data using a target-perturbed gene regulatory networkLi, Chengyu et al.Cell Reports Medicine, Volume 5, Issue 6, 101568

Researchers developed BsPNEC, a bispecific peptide-nanozyme conjugate targeting PD-L1 and CXCR1/2. It disrupts immunosuppression, activates the cGAS-STING pathway via ROS to boost T cell infiltration, and enables MRI. It effectively treats resistant pancreatic cancer.

References:

• Chen D, Xu R, Ye X, et al. Design of a bispecific peptide-nanozyme conjugate for cancer immunotherapy[J]. Cell Reports Medicine, 2026.

Researchers used spatial proteomics on 2,019 tumors to identify five TIME spatial factors conserved across cancer types. These factors, like immune enrichment and mixing, link to genomic alterations and patient survival. A 39.4-million-cell single-cell database was also created.

References:

• Pan-cancer spatial characterization of key immune biomarkers in the tumor microenvironmentLindsay, James R. et al.Cell Reports Medicine, Volume 0, Issue 0, 102418

The FOCUS method is an AI-assisted framework designed to manage information overload in research. By using tools like NotebookLM and Gemini, scholars can find, organize, condense, understand, and synthesize vast data while maintaining intellectual rigor and privacy.

References:

• Lin Z. FOCUS: an AI-assisted reading workflow for information overload: Career feature[J]. Nature Biotechnology, 2025: 1-6.

This study benchmarks 27 computational methods for predicting single-cell perturbation responses across 29 datasets. It reveals that no single algorithm excels in all tasks. Foundation models improve with large data, but all methods struggle with cellular context generalization.

References:

• Wei, Z., Wang, Y., Gao, Y. et al. Benchmarking algorithms for generalizable single-cell perturbation response prediction. Nat Methods (2025). doi.org

Researchers developed DrugCLIP, an AI tool for genome-wide virtual screening. Using deep contrastive learning, it screens trillions of pairs in under 24 hours. Validated by wet-lab experiments on targets like TRIP12, it offers an open-access database for drug discovery.

References:

• Jia Y, Gao B, Tan J, et al. Deep contrastive learning enables genome-wide virtual screening[J]. Science, 2026, 391(6781): eads9530.