Journal: NAR Genomics and Bioinformatics

Article Title: SpNeigh: spatial neighborhood and differential expression analysis for high-resolution spatial transcriptomics

doi: 10.1093/nargab/lqag039

Figure Lengend Snippet: Neighborhood analysis reveals immune cell distribution differences around tumor and DCIS regions. ( a ) Spatial plot of Xenium breast cancer tissue colored by cell type annotations. Neighbor ring regions surrounding tumor and DCIS are shown separately for clarity. Blue indicates tumor-associated rings; yellow indicates DCIS-associated rings. ( b ) Cells located within tumor and DCIS neighbor rings. ( c ) Overall cell type proportions in tumor and DCIS neighbor rings. Cells from all rings are aggregated for each condition to compute relative frequencies (labels shown for proportions \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $\ge$\end{document} 2%). ( d ) Row-scaled spatial interaction matrices showing the frequency of neighboring cell types around each focal cell type within tumor and DCIS neighborhoods. ( e ) Spatial distribution of major immune cell types, including T cells, macrophages, plasma cells, and B cells. Blue and yellow polygons indicate boundaries of tumor and DCIS, respectively. Notably, T, B, and plasma cells appear more enriched near DCIS than tumor regions. ( f ) Bar plot comparing T cell proportions between tumor and DCIS neighbor rings. Blue bars represent tumor-associated neighborhoods and yellow bars represent DCIS. The T cell proportion is significantly higher near DCIS (Student’s t -test, P = .012).

Article Snippet: Human breast cancer Xenium dataset: https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast [ ].

Techniques: Clinical Proteomics

Journal: bioRxiv

Article Title: Reconstructing biologically coherent cellular profiles from imaging-based spatial transcriptomics

doi: 10.64898/2026.03.08.710395

Figure Lengend Snippet: (A) Global gene–gene NPMI matrix computed from the standard Xenium breast cancer dataset, showing block-diagonal lineage programs and negative cross-lineage associations. Insets (black boxes) correspond to panels B–C. (B,C) Representative lymphoid (B) and myeloid (C) gene modules with uniformly high within-lineage NPMI. (D) UMAP embedding of whole-cell profiles colored by the NPMI-derived purity score, which quantifies internal gene–gene coherence. Cells in the top 20% of purity scores are indicated, marking transcriptionally coherent entities. (E) Distribution of purity scores across cells with a two-component Gaussian mixture model (GMM), revealing a bimodal structure separating high-purity from low-purity profiles. (F) UMAP embedding colored by the NPMI-derived conflict score, which captures mutually incompatible gene associations within individual cells. Cells in the top 20% of conflict scores localize predominantly to regions connecting distinct UMAP clusters. (G) Distribution of conflict scores and corresponding two-component GMM fit, separating low-conflict and high-conflict states associated with mixed-lineage transcriptional signatures.

Article Snippet: The standard Xenium breast cancer dataset can be accessed via https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast .

Techniques: Blocking Assay, Derivative Assay

Journal: bioRxiv

Article Title: Reconstructing biologically coherent cellular profiles from imaging-based spatial transcriptomics

doi: 10.64898/2026.03.08.710395

Figure Lengend Snippet: (A) UMAP embedding of whole-cell profiles from a multimodal Xenium (MX) lung cancer dataset colored by the NPMI-derived purity score. Cells within the top 20th percentile of the purity distribution are highlighted. (B) UMAP embedding colored by the NPMI-derived conflict score, with the top 20th percentile of high-conflict cells highlighted. (C,D) Distributions of purity (C) and conflict (D) scores fitted using two-component Gaussian mixture models (GMMs). (E,F) UMAP embeddings and corresponding marker-gene dot plots for TRACER refinements applied to the multimodal Xenium segmentation. (E) TRACER-stitched refinement. (F) TRACER-fine-tuned refinement. (G) Assignment reproducibility across two independent TRACER runs with different random seeds (0 and 42), quantified using exact-match fraction, adjusted Rand index (ARI), and normalized mutual information (NMI). TRACER assignments are identical across runs (exact-match fraction = 1.0). (H) Per-cell changes in relative purity and relative conflict across the two runs, with refinement-associated shifts toward higher purity and lower conflict at the per-cell level.

Article Snippet: The standard Xenium breast cancer dataset can be accessed via https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast .

Techniques: Derivative Assay, Marker

Journal: bioRxiv

Article Title: Reconstructing biologically coherent cellular profiles from imaging-based spatial transcriptomics

doi: 10.64898/2026.03.08.710395

Figure Lengend Snippet: (A) Three-dimensional concave hull constructed from the transcript point cloud originally assigned to a representative T cell–tumor hybrid cell under standard Xenium segmentation. (B,C) TRACER-extracted component corresponding to the tumor-associated subset of transcripts from the hybrid cell: (B) 3D concave hull and (C) transcript frequency profile. (D,E) TRACER-extracted component corresponding to the T cell–associated subset of transcripts: (D) 3D concave hull and (E) transcript frequency profile. (F) UMAP embedding of partial cells identified by TRACER after stitching and spatial fine-tuning. (G) Differential marker-gene expression matrix for Leiden clusters (annotated by marker-gene expression) computed from the partial cell population. (H) Spatial distribution of whole cells in the breast cancer Xenium section, colored by the original cell-type annotations. (I) Spatial distribution of TRACER-derived partial cells originating from segmentation-derived hybrid profiles, colored by Leiden clusters. (J) Spatial distribution of TRACER-derived partial cells reconstructed from previously unassigned transcripts, colored by Leiden clusters. The left inset shows transcripts contributing to reconstructed partial cells (yellow) overlaid on standard Xenium whole-cell segmentation boundaries (red). The right inset shows a convex-hull representation of these reconstructed partial cell transcripts within extracellular regions.

Article Snippet: The standard Xenium breast cancer dataset can be accessed via https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast .

Techniques: Construct, Marker, Gene Expression, Derivative Assay

Journal: bioRxiv

Article Title: Reconstructing biologically coherent cellular profiles from imaging-based spatial transcriptomics

doi: 10.64898/2026.03.08.710395

Figure Lengend Snippet: (A) Three-dimensional concave hull constructed from the transcript point cloud originally assigned to an unlabeled cell under the standard Xenium segmentation. This cell displayed promiscuous marker gene expression, preventing confident assignment in the original study. Three orthogonal views of the reconstructed concave hull are shown. (B,C) TRACER-derived component corresponding to a T cell, represented as a 3D concave hull (B) together with its gene frequency profile (C). (D,E) TRACER-derived component corresponding to a tumor cell, represented as a 3D concave hull (D) and its gene frequency profile (E). (F) Differential marker gene expression matrix for cell-type annotations defined by the original study applied to the standard Xenium segmentation, showing mixed-lineage signal across annotated groups. (G) Differential marker gene expression matrix for the same published cell-type annotations applied to TRACER-stitched correction, showing more distinct lineage-associated marker expression. Cell-type labels are derived entirely from the original peer-reviewed dataset and were not used during TRACER refinement.

Article Snippet: The standard Xenium breast cancer dataset can be accessed via https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast .

Techniques: Construct, Marker, Gene Expression, Derivative Assay, Expressing