Research Radar — 2026-07-19
Methods & AI
Computational
Whole-transcriptome-scale isoform-resolved spatial imaging of single cells in tissues
Cell Published 2026-07-15 spatial_omics DOI: 10.1016/j.cell.2026.07.015
spatial transcriptomics single-cell isoform resolution imaging MERFISH
Summary: RT&T-AMP-MERFISH extends the MERFISH spatial imaging platform to achieve whole-transcriptome-scale, isoform-resolved detection of RNA molecules in single cells within intact tissues. The method combines targeted and untargeted amplification strategies to simultaneously profile thousands of genes at single-cell resolution, revealing how gene programs and transcript isoform usage vary across cell types and anatomical regions in the brain.
Why it matters: Represents a landmark technical achievement in spatial transcriptomics, pushing the field toward true whole-transcriptome coverage with isoform resolution in intact tissues — a capability that will fundamentally transform how we study tissue organization, development, and disease at the RNA level.
Why for Yiru: Directly relevant to my spatial transcriptomics research — this whole-transcriptome-scale approach could be applied to characterize the tumor microenvironment at unprecedented depth, revealing alternative isoform usage in immune cells that may inform immunotherapy response and resistance mechanisms.
Spatial proximity sequencing maps developmental dynamics in the germinal center
Cell Published 2026-07-15 spatial_omics DOI: 10.1016/j.cell.2026.07.034
spatial transcriptomics proteomics germinal center B cell protein complexes
Summary: Sprox-seq is a spatial proximity sequencing method that simultaneously profiles protein complexes, surface proteins, and mRNAs in intact tissues. Applied to human tonsils, it maps germinal center interaction networks, links CD21-CD35 complexes to proliferative programs, reveals interaction-based B cell state transitions, and directly captures B cell-follicular dendritic cell communication.
Why it matters: Bridges a critical gap between spatial transcriptomics and proteomics by capturing functional protein-level information — including protein complexes and surface protein interactions — within the spatial tissue context, adding a functional dimension to spatial omics that transcript abundance alone cannot provide.
Why for Yiru: The integration of protein-level spatial information with transcriptomics is highly relevant to my tumor microenvironment research, where surface protein expression on immune cells and their spatial interactions define functional states and therapeutic targets in immunotherapy.
SpaDiff denoises sequence-based spatial transcriptomics via diffusion process
Cell Reports Methods Published 2026-07-16 computational_method DOI: 10.1016/j.crmeth.2026.07.016
spatial transcriptomics denoising diffusion model computational biology deep learning
Summary: Sequence-based spatial transcriptomics suffers from spot-swapping, where RNA molecules drift to neighboring spots and blur tissue patterns. Cai et al. introduce SpaDiff, a diffusion-based denoiser that reverses this spreading along a learned spatial gradient, sharpening tissue domains while preserving total molecular counts exactly and recovering within-region heterogeneity that other methods erase.
Why it matters: Addresses a critical and underappreciated technical limitation in sequencing-based spatial transcriptomics — spot-swapping — with a principled diffusion-based approach that preserves quantitative accuracy, improving the reliability of downstream analyses in spatial biology.
Why for Yiru: Directly relevant to my spatial transcriptomics analyses — denoising spot-swapping artifacts with SpaDiff would improve the resolution of tissue domain boundaries and cell-cell interaction inference in my tumor microenvironment studies.
Foundation model reveals the shared organization of transcription and topologically associating domains
Cell Systems Published 2026-07-16 foundation_model DOI: 10.1016/j.cels.2026.07.016
foundation model chromatin TAD gene regulation epigenomics
Summary: Liang et al. develop a species-level consensus TAD Map paired with foundation model-derived contextual similarity to reveal a systematic 20% transcriptional enhancement within TADs across all genomic distances. This organization strengthens during development, weakens with aging, and shifts in cancer cells, suggesting a probabilistic chromatin regulatory mechanism governing transcription.
Why it matters: Provides the first species-level consensus map of topologically associating domains and demonstrates with a foundation model that TADs systematically enhance transcription, revealing a fundamental principle of genome organization that is dynamically regulated across biological states.
Why for Yiru: The foundation model approach to characterizing chromatin organization is relevant to my broader interest in deep learning for biology, and the TAD biology may inform how spatial genome organization impacts gene expression programs in immune cells within the tumor microenvironment.
AI-enabled reconstruction of 3D spatial multi-omics at single-cell resolution
bioRxiv (Bioinformatics) Published 2026-07-13 spatial_omics DOI: 10.64898/2026.07.09.737490v1
3D reconstruction spatial multi-omics deep learning tumor microenvironment computational biology
Summary: Histo3D-MO is a hybrid experimental-computational pipeline that reconstructs single-cell-resolution 3D spatial multi-omics maps by integrating sparse, omics-disjoint spatial measurements with dense H&E histology through SPONGE, a deep learning model. Applied to hepatocellular carcinoma, it reveals spatially organized translation efficiency patterns, volumetric decoupling between malignant cells and monocytes, and depth-associated monocyte differentiation trajectories.
Why it matters: Establishes a scalable framework for single-cell-resolution 3D spatial multi-omics that overcomes the prohibitive cost of direct 3D molecular profiling, enabling whole-volume characterization of the tumor microenvironment and complex biological systems.
Why for Yiru: Highly relevant to my spatial multi-omics and tumor microenvironment research — the ability to reconstruct 3D molecular maps from 2D sections using deep learning opens new possibilities for understanding spatial organization in the TME, particularly depth-dependent immune cell differentiation and tumor-immune interactions.
Single cell multi-omics enables high-resolution identification and functional purification of human acute myeloid leukemia stem cells
bioRxiv (Cancer Biology) Published 2026-07-12 single_cell DOI: 10.64898/2026.07.12.737989v1
single-cell multi-omics AML leukemia stem cells transcriptomics
Summary: This study analyzes large patient cohorts with bulk gene expression and single-cell multi-omic assays to identify a prognostic gene signature specifically enriched in a clinically adverse AML sub-population. Using this signature, the authors define and prospectively isolate CD34+CD90-CLL1-CD69+CD53- immunophenotypic leukemia stem cells that are significantly enriched for LSC content by limiting dilution xenotransplantation.
Why it matters: Demonstrates the power of single-cell multi-omics to precisely identify clinically relevant stem cell populations that have been inaccessible to prospective isolation, establishing a clear framework for translating single-cell signatures into immunophenotypic definitions for functional studies and therapeutic targeting.
Why for Yiru: Directly relevant to my single-cell multi-omics work — the approach of translating single-cell signatures into flow-sortable immunophenotypes provides a template for isolating rare cell populations in the tumor microenvironment, including cancer stem cells and progenitor-like T cells.
Biomedical discoveries
Biomedicine
GPC3-specific dnTGFβRII-armoured CAR T cells for hepatocellular carcinoma
Nature Published 2026-07-14 immunotherapy DOI: 10.1038/s41586-026-10786-z
CAR-T immunotherapy hepatocellular carcinoma TGF-beta solid tumors
Summary: Chimeric antigen receptor (CAR) T cells targeting GPC3 are 'armoured' with a dominant-negative TGFβ receptor II to render them resistant to the immunosuppressive tumor microenvironment. The armoured CAR T cells show enhanced efficacy and tolerable safety in patients with advanced hepatocellular carcinoma who have developed resistance or not responded to previous treatment.
Why it matters: Provides clinical proof-of-concept for armoured CAR T cell designs that neutralize TGFβ-mediated immunosuppression in solid tumors, addressing one of the major barriers to CAR T cell efficacy in the hostile solid tumor microenvironment.
Why for Yiru: Directly relevant to my CAR-T and immunotherapy research — the dominant-negative TGFβRII armouring strategy is a design principle I can explore for other solid tumor targets, and the clinical validation in HCC establishes the feasibility of engineering CAR T cells to resist TME suppression.
A targetable micropeptide reprograms macrophages to suppress T cell anti-tumor immunity
Nature Cancer Published 2026-07-14 cancer_biology DOI: 10.1038/s43018-026-01196-1
macrophage micropeptide cGAS-STING tumor microenvironment immunotherapy
Summary: This study identifies UEIS, a micropeptide expressed in tumor-associated macrophages that suppresses anti-tumor immunity by dampening cGAS-STING type I interferon signaling. UEIS sequesters TBK1 within biomolecular condensates in a negative feedback loop, and targeted disruption of UEIS condensation restores anti-tumor responses and improves the efficacy of immune checkpoint blockade in both human cells and mouse models.
Why it matters: Reveals a previously unknown micropeptide-based immune checkpoint in macrophages that operates through phase separation mechanisms, opening a new class of therapeutic targets — micropeptides that regulate innate immune signaling via biomolecular condensation.
Why for Yiru: Highly relevant to my macrophage biology and tumor microenvironment research — the UEIS-TBK1 axis represents a new mechanism of macrophage immunosuppression that could be targeted therapeutically, and the micropeptide paradigm may extend to other innate immune cells in the TME.
PD-1 blockade unleashes local hepatitis B virus-related B cell response inhibiting hepatocellular carcinoma
Cancer Cell Published 2026-07-15 translational_research DOI: 10.1016/j.ccell.2026.07.015
immunotherapy B cell hepatocellular carcinoma HBV spatial transcriptomics
Summary: Following perioperative anti-PD-1 therapy in hepatocellular carcinoma, Chen et al. identify two distinct subtypes of late-recurrence patients characterized by either T cell or B cell dominant responses within tumors. Using single-cell spatial transcriptomics and antibody cloning, they show that anti-PD-1 induces local HBV-targeted antibody responses and complement-mediated anti-tumor activity.
Why it matters: Reveals that B cell responses are a major and previously underappreciated mechanism of anti-PD-1 efficacy in virus-associated cancers, shifting the paradigm beyond T cell-centric views of checkpoint blockade and highlighting HBV-targeted humoral immunity as a key therapeutic axis.
Why for Yiru: Directly relevant to my immunotherapy and tumor microenvironment research — the integration of spatial transcriptomics with B cell repertoire analysis provides a framework for studying humoral immunity in the TME, and the B cell mechanism expands my understanding of how checkpoint blockade works in virus-associated cancers.
MEK-dependent bioenergetic demand drives terminal CD8+ T cell exhaustion
Immunity Published 2026-07-12 immunology DOI: 10.1016/j.immuni.2026.07.012
T cell exhaustion metabolism MEK immunotherapy
Summary: Mitra et al. examine how chronic antigen stimulation of CD8+ T cells leads to metabolic dysfunction and find that MEK-driven ATP demand drives terminal T cell exhaustion by supporting protein synthesis at the expense of NAD+-dependent proliferation. MEK inhibition restores proliferative capacity and maintains a progenitor-like exhausted T cell state that can respond to immune checkpoint blockade.
Why it matters: Provides a mechanistic framework linking MEK signaling to T cell exhaustion through bioenergetic competition between protein synthesis and NAD+ metabolism, identifying MEK inhibition as a strategy to preserve T cell functionality during chronic antigen exposure.
Why for Yiru: Relevant to my immunotherapy research — understanding how metabolic programs drive T cell exhaustion suggests combination strategies (MEK inhibition + checkpoint blockade) that could improve CAR-T cell persistence and activity in the metabolically hostile tumor microenvironment.
The transcription factor BACH1 couples chromatin priming and repression to enable macrophage plasticity and adaptation
Immunity Published 2026-07-16 immunology DOI: 10.1016/j.immuni.2026.07.016
macrophage transcription factor chromatin plasticity epigenetics
Summary: Tzerpos et al. find that the transcription factor BACH1 establishes early chromatin accessibility while maintaining transcriptionally repressive states in naive macrophages before activation. This dual activity — termed pioneer repression — regulates macrophage adaptation and inflammatory kinetics in vivo, enabling rapid yet controlled transcriptional responses to environmental cues.
Why it matters: Identifies a novel paradigm in transcription factor biology — pioneer repression — where a single factor simultaneously primes chromatin for future activation while maintaining repression, explaining how macrophages achieve both rapid responsiveness and precise transcriptional control.
Why for Yiru: Directly relevant to my macrophage biology research — BACH1's dual pioneer-repressor function provides a new framework for understanding how tumor-associated macrophages acquire and maintain their immunosuppressive programs, with implications for targeting macrophage plasticity in cancer immunotherapy.
Single-cell transcriptomic unveils tumor-mediated reprogramming of neutrophils and their unique vulnerability that inhibits metastasis
Science Advances Published 2026-07-15 cancer_biology DOI: 10.1126/sciadv.aee9308
neutrophil metastasis tumor microenvironment single-cell immunosuppression
Summary: This study uses single-cell transcriptomics to characterize how tumors reprogram neutrophils, revealing distinct neutrophil states within the tumor microenvironment. The analysis identifies a unique vulnerability in metastasis-associated neutrophils that can be therapeutically targeted to inhibit metastatic spread without compromising systemic neutrophil function.
Why it matters: Provides a high-resolution map of neutrophil heterogeneity in the TME and identifies metastasis-specific neutrophil programs as therapeutic targets, addressing the growing recognition of neutrophils as key regulators of metastasis.
Why for Yiru: Relevant to my tumor microenvironment interests — understanding neutrophil reprogramming in metastasis complements my work on macrophage biology and myeloid cell heterogeneity, and the identified vulnerabilities could inform combination immunotherapy strategies targeting the myeloid compartment.
Cross-disciplinary watchlist
Other Fields
Spatio-DARLIN maps cell state and clonal history in intact mouse tissues
Nature Methods Published 2026-07-16 spatial_omics DOI: 10.1038/s41592-026-03157-z
lineage tracing spatial transcriptomics clonal architecture development genomics
Summary: Spatio-DARLIN combines high-diversity genetic lineage tracing with spatial transcriptomics to simultaneously record cell lineage, gene expression, and spatial location in the same tissue section. This approach reliably recovers thousands of clones in mouse intestine and brain, revealing how clonal architecture delineates tissue development and homeostasis.
Why it matters: Integrates two powerful technologies — lineage tracing and spatial transcriptomics — into a single platform, enabling the direct linking of clonal history to spatial organization and cell state, which is essential for understanding development, regeneration, and tumor evolution.
Why for Yiru: Highly relevant to my spatial transcriptomics research — Spatio-DARLIN could be applied to trace the clonal origins of tumor subclones and immune cell populations within the tumor microenvironment, revealing how spatially organized clonal lineages drive tumor progression and immune evasion.
Large-scale multi-sequence pretraining for generalizable MRI analysis in versatile clinical applications
Nature Biomedical Engineering Published 2026-07-12 foundation_model DOI: 10.1038/s41551-026-01740-5
foundation model MRI deep learning biomedical AI medical imaging
Summary: MARS is a foundation model for magnetic resonance imaging trained with large-scale multi-sequence pretraining that overcomes limitations in image heterogeneity. It enables versatile real-world multi-sequence MRI analysis on diverse anatomical structures, generalizing across different imaging protocols and body regions without task-specific fine-tuning.
Why it matters: Demonstrates that a foundation model pretrained on diverse multi-sequence MRI data can generalize across anatomical structures and imaging protocols, establishing a new paradigm for medical image analysis that reduces the need for task-specific models.
Why for Yiru: Relevant to my broader interest in foundation models for biomedical applications — the multi-sequence pretraining strategy and the importance of data diversity for generalization are principles I can apply when designing foundation models for spatial transcriptomics and single-cell data.
A Bayesian framework for longitudinal EHR and genetic discovery
Nature Published 2026-07-14 biomedical_ai DOI: 10.1038/s41586-026-10780-5
bayesian EHR genetics disease discovery computational biology
Summary: A Bayesian generative framework that integrates longitudinal electronic health records with genetic data is presented, enabling the identification of latent disease signatures that capture both temporal disease progression and genetic predisposition. The framework models disease as a continuous latent process informed by both clinical measurements and genomic variation.
Why it matters: Provides a principled statistical approach to integrating longitudinal clinical data with genomics, moving beyond cross-sectional GWAS to capture the temporal dynamics of disease onset and progression in a unified Bayesian model.
Why for Yiru: While outside my primary focus on spatial and single-cell omics, the Bayesian integration framework offers methodological inspiration for combining multi-modal biological data — a challenge I face in integrating spatial transcriptomics with clinical outcomes and genomic data.
An encyclopedia of human enhancer-gene regulatory interactions
Nature Published 2026-07-14 genomics DOI: 10.1038/s41586-026-10781-4
enhancer gene regulation encyclopedia genomics regulatory interactions
Summary: This study presents a comprehensive encyclopedia of human enhancer-gene regulatory interactions, systematically mapping connections between distal regulatory elements and their target genes across multiple cell types and tissues. The resource provides a foundational reference for interpreting non-coding genetic variation and understanding gene regulatory mechanisms.
Why it matters: Provides a systematic reference map of enhancer-gene interactions that will serve as a critical resource for interpreting genome-wide association study loci, understanding regulatory mechanisms in development and disease, and prioritizing non-coding variants for functional validation.
Why for Yiru: Useful as a reference resource for my work on gene regulation — the enhancer-gene interaction map can help interpret regulatory changes I observe in spatial transcriptomics data and single-cell ATAC-seq analyses of the tumor microenvironment.