Wu Wei

Translating genetic risk for complex traits into mechanistic insight requires locating risk signals within native tissue architectures. Although spatial transcriptomics (ST) provides high-resolution anatomical context, its integration with genome-wide association studies (GWAS) can be confounded by linkage disequilibrium (LD) and spatial autocorrelation. These factors can generate spurious associations and spread localized genetic signals across non-specific tissue regions. Here we establish a genotype-aware spatial simulation benchmark and develop SDESE (Spatially refined Driver Estimation by Selective Expression), a framework for joint spatial and gene fine-mapping of complex phenotypes. SDESE couples spatial localization with gene prioritization through iterative conditional association analysis, thereby reducing signals attributable to LD proxies and co-expression. We also provide a multi-organ spatial resource and analysis platform for downstream applications. In spatial simulations, SDESE retained high sensitivity while reducing non-specific spatial spread, particularly for localized signals. We then applied SDESE to adult human dorsolateral prefrontal cortex (DLPFC) ST data from five common psychiatric disorders and validated the results using multiple independent GWAS datasets. Genetic risk across these disorders converged on shared neuroanatomical niches, particularly mid-to-deep cortical layers and excitatory neuronal microenvironments. However, the corresponding associated-gene sets showed limited overlap, while functional enrichment implicated neurodevelopmental, synaptic and ion-transport processes across disorders. These findings reveal a layered spatial-genetic architecture in which shared neuroanatomical vulnerability coexists with partially distinct but functionally convergent sets of associated genes.

The principles organizing cellular diversity and connectivity in primate brains remain elusive. By integrating spatial transcriptomics, magnetic resonance imaging, and retrograde labeling in marmosets, we identified two opposing molecular gradients that undergo postnatal refinement, emanating from allocortices and primary sensory cortices, respectively. These gradients reconcile conflicting hypotheses on cortical expansion and characterize distinct cortical areas. Cortical gradients align with thalamic gene expression and thalamocortical projection patterns. At gradient intersections, the default mode network and frontal pole exhibited similar molecular features in humans and marmosets, despite species-specific differences in functional connectivity. Comparative analysis of gradient-related genes showed that marmoset and human auditory cortices are highly similar but differ from those of macaques, potentially reflecting complex vocalization. Together, these opposing gradients represent a fundamental organizing principle of the primate cortex.

Accurate cell instance segmentation is foundational for digital pathology analysis. Existing methods based on contour detection and distance mapping still face significant challenges in processing complex and dense cellular regions. Graph coloring-based methods provide a new paradigm for this task, yet the effectiveness of this paradigm in real-world scenarios with dense overlaps and complex topologies has not been verified. Addressing this issue, we release a large-scale dataset GBC-FS 2025, which contains highly complex and dense sub-cellular nuclear arrangements. We conduct the first systematic analysis of the chromatic properties of cell adjacency graphs across four diverse datasets and reveal an important discovery: most real-world cell graphs are non-bipartite, with a high prevalence of odd-length cycles (predominantly triangles). This makes simple 2-coloring theory insufficient for handling complex tissues, while higher-chromaticity models would cause representational redundancy and optimization difficulties. Building on this observation of complex real-world contexts, we propose Disco (Densely-overlapping Cell Instance Segmentation via Adjacency-aware COllaborative Coloring), an adjacency-aware framework based on the "divide and conquer" principle. It uniquely combines a data-driven topological labeling strategy with a constrained deep learning system to resolve complex adjacency conflicts. First, "Explicit Marking" strategy transforms the topological challenge into a learnable classification task by recursively decomposing the cell graph and isolating a "conflict set." Second, "Implicit Disambiguation" mechanism resolves ambiguities in conflict regions by enforcing feature dissimilarity between different instances, enabling the model to learn separable feature representations.

Liver invasion is one of the most frequent events in the progression of gallbladder cancer (GBC). However, the cellular and pathological role of the tumor-liver–interface microenvironment in liver invasion is still enigmatic. Here, we applied single-cell and spatial transcriptomics to systematically investigate the cellular component and gene expression regulation of the microenvironment from the tumor to the liver, specifically the invasive boundary. Our analyses revealed that CXCL9⁺ macrophage–rich immune cell niches were accumulated in the tumor-liver invasive margin, where 2 subclasses of the CXCL9⁺ immune cell niches, CXCL9⁺TRAC⁺ (CT) and CXCL9⁺C1QB⁺ (CC) niches, were identified. CD8⁺ T cells were recruited by CXCL9⁺ macrophages through CXCL9-CXCR3 interaction in the CT niche, which was located adjacent to the liver. Moreover, the CC niche was proximal to the tumor core, where tumor cells induced CD8⁺ T cell exhaustion via LGALS4 expression. In addition, our cohort study showed that high CXCL9 and low LGALS4 in the liver invasion margin demonstrated a favorable prognosis and better responses to anti–PD-1 immunotherapy for patients with gallbladder cancer. Altogether, these findings demonstrate novel cellular and molecular mechanisms underlying liver invasion and offer clinical value for immunotherapies.

Supplementary Table S7. Patient information.

Supplementary Table S2. Differential ASEs, primers and sequences information.

Supplementary Table S1. Samples and transcriptomics information.

Supplementary Table S6. GO term and KEGG pathway information in Gzmb+ Cd8+ T cells.