Biology subjects

Birk, S.

Publications and source records attributed to Birk, S..

8 recordsLinked to original sources

bioRxiv2026

Fine-Grained and Reliable Cell Classification from Pathology Images

Molecular assays such as spatial transcriptomics facilitate fine-grained identification of cell types for analyzing tumor microenvironments. However, they are expensive to acquire in routine clinical practice compared to hematoxylin and eosin (H&E) images. With molecular guidance, such images can distinguish cell lineages, yet often fail to resolve closely related subtypes. To overcome this, we introduce CellBridge, a multi-modal contrastive learning framework pretrained on 16M cells with paired H&E and spatial profiles, for transferring molecular information into morphological representations. CellBridge further incorporates a Cell-Ontology-aware conformal prediction module that assigns labels at different levels of specificity: leaf-level cell types when morphology alone provides sufficient evidence, and broader ancestor categories when distinguishing among more specific cell types would require molecular information. Across in-domain and out-of-domain benchmarks, CellBridge outperforms leading pathology foundation models in macro F1 by up to 16 percentage points, with the largest gains on the most fine-grained label spaces. It consistently achieves smaller prediction sets, while reducing class-wise coverage gaps. We envision CellBridge as a step toward extracting cellular information that currently requires costly molecular assays from routinely acquired pathology images, while explicitly identifying where such assays remain necessary. Our code is available at https://github.com/Lotfollahi-lab/cellbridge.

bioinformatics↗

bioRxiv2026

G2T: Tissue Reconstruction from Gene Expression via Embedding-Distance Flow Matching

Single-cell RNA sequencing (scRNA-seq) profiles transcriptomes at high resolution but discards the spatial context of cells within a tissue -- information that is essential for studying intercellular mechanisms and tissue architecture. Spatial transcriptomics (ST) retains coordinates but, depending on the assay, trades this off against gene-panel breadth, spatial resolution, or cost. We present G2T (Gene-to-Tissue), a generative deep learning model that reassembles a tissue from gene expression -- its only observed input -- by predicting the matrix of pairwise distances between cells in a learned embedding space. G2T uses an attention-based Transformer with an Euclidean-Distance-Matrix (EDM) output head and is trained with conditional flow matching: the network learns to denoise corrupted cell positions, conditioned on the slice's gene expression, by predicting per-cell embeddings whose pairwise squared distances match the ground-truth distance matrix. At inference, a fast locally-optimal-block (LOBPCG) multidimensional scaling step turns the predicted distance matrix into 2-D coordinates. On a published MERFISH mouse primary motor cortex benchmark, G2T improves over the previous state-of-the-art method, LUNA, across all three standard metrics -- Spearman correlation of pairwise-distance ranks, Contact F1, and per-cell-class Sum RSSD -- and even larger relative gains on the mouse central-nervous-system scRNA-seq atlas, evaluated against an imputed spatial reference (STARmap PLUS-integrated locations, not measured coordinates). By predicting this geometry in a higher-dimensional embedding space rather than regressing 2-D coordinates, G2T relaxes the 2-D output parameterisation of prior diffusion-based methods and yields a compact, scalable building block for reconstructing tissue from dissociated cells, enabling downstream spatial niche and cell-cell communication analysis.

genomics↗

bioRxiv2026

Learning Discrete Cell and Niche Codes from Spatial Transcriptomics Using Dual Residual Vector Quantization

Spatially-resolved transcriptomics (SRT) measures gene expression at single-cell resolution while preserving each cells spatial location, enabling the joint study of cell identity and cellular niche, the recurring microenvironment that organises tissue function. Existing representation-learning methods typically capture only one of these axes at a time. We present SQUINT, a graph vector-quantized variational autoencoder (VQ-VAE) that learns two disjoint codebooks per cell from a shared architecture: a cell codebook quantising the per-cell embedding before neighbourhood aggregation, biased toward cell-intrinsic identity, and a niche codebook quantising the embedding after graph neural network (GNN) aggregation, biased toward spatial context. Both use residual vector quantization, giving a coarse-to-fine discrete-token hierarchy. SQUINT is trained with per-branch negative-binomial reconstruction objectives and three domain-motivated components that we show are crucial: a within-section cosine adjacency loss that anchors the niche codes in the spatial graph, a cross-section contrastive loss on the cell latents that aligns transcriptomically matched cells, and a decoder section covariate that absorbs batch effects. Across three datasets spanning four spatial assays (STARmap, MERFISH, CosMx, Xenium) and four tasks - niche identification, cell-type identification, cross-section integration, and spatial gene-expression imputation in held-out regions - SQUINT outperforms or is competitive with strong baselines on identification and achieves the most faithful cross-section integration. The resulting discrete vocabulary makes tissues directly consumable by transformer-style foundation models and enables one-step query-to-reference atlas mapping via code-distribution similarity, which we demonstrate on a CosMx human non-small-cell lung cancer cohort.

genomics↗

bioRxiv2026

Multi-scale modeling of human tissues from spatial transcriptomics with TERRA

Spatial transcriptomics maps gene expression at cellular resolution, revealing how cells organize into multicellular niches. Yet computational analyses remain dataset-specific, without a transferable representation of tissue organization that generalizes across datasets, tasks and tissues or predicts how tissues behave under perturbation. We present TERRA, a foundation model pretrained on 112 million human cells profiled by spatial transcriptomics. From a single pretrained backbone, TERRA yields embeddings at the scale of cells, the genes they express and the neighborhoods in which they reside, and supports spatial in silico perturbation, all applied zero-shot to unseen tissues. At the cell level, in newly generated spatial data for developing pancreas, TERRA identified an islet-associated capillary state which we posit represents a developmental precursor of the mature islet microvasculature. At the gene level, in untreated kidney sections, in silico knockout of immune-checkpoint targets predicted a gene program of immune-checkpoint-blockade-associated nephrotoxicity, which we validated in treatment-exposed tissue and recovered in blood. At the neighborhood level, TERRA mapped macrophages across tissues to identify recurring cross-organ niches, which we term archetypes, including a tumor-boundary niche associated with poor prognosis in kidney cancer. Together, TERRA captures the spatial and multicellular logic of human tissue and predicts, in silico, its response to perturbation, providing a multi-scale framework for tissue biology, therapeutic development and clinical application.

genomics↗

bioRxiv2026

Hidden immune memory niches in inflammatory skin diseases

Disease-associated histopathological features are widely used to identify tissue microenvironments or niches for diagnostics and treatment response in clinical practice. However, despite its widespread use, histopathology does not reveal the full cellular and molecular composition of known pathological niches. Furthermore, the existence of pathological niches that may not be histologically discernible remains unknown. In this study, we generated a spatially-resolved multi-modal molecular atlas of [~]5 million human skin cells (including 113 skin sections profiled using Xenium-5k) and applied deep learning to unbiasedly decode 26 skin niches in health and disease. Several disease-associated niches corresponded to known histopathological features, and we defined their cellular and molecular features, co-localisations, and interactions. Additionally, we discovered an immunologically active role for skin appendageal structures in disease mechanisms, potentially contributing to inflammatory memory, that was not identifiable using standard histopathological analysis. These include a resident memory T cell-rich niche in the sebaceous gland and a plasma cell-rich niche in the sweat gland, analogous to the gland-associated immune niche in lung. Finally, we illustrate how our atlas can be used to generate high-resolution representations using transfer learning, resolving rare T cell and sebocyte subsets not possible in the original studies, validating niche identification, and the spatial enrichment of candidate genes linked to disease-associated genetic variants. Overall, our study links histopathology and atlas-scale genomics to reveal novel insights into inflammatory disease pathogenesis, chronicity, and potentially curative therapeutic avenues, using skin as an exemplar tissue for this approach.

immunology↗

bioRxiv2025

Mapping and reprogramming microenvironment-induced cell states in human disease using generative AI

Tissue microenvironments reprogram local cellular states in disease, yet current computational spatial methods remain descriptive and do not simulate tissue perturbation. We present MintFlow, a generative AI algorithm that learns how the tissue microenvironment influences cell states and predicts how tissue perturbations can reprogram them. Applied to three human diseases, MintFlow uncovered distinct pathogenic spatial reprogramming in inflammatory and tumor microenvironments. In atopic dermatitis, MintFlow identified a novel, spatially-imprinted, type 2 (IL13+ITGAE+) epidermal T resident memory cell population (type 2 TRM), and decoded signaling pathways within the perivascular lymphoid niche. In melanoma, MintFlow identified fibrotic stroma resembling keloid scar tissue. In kidney cancer, MintFlow resolved immunosuppressed CD8+ T cell states within tertiary lymphoid structures. Furthermore, MintFlow enabled in silico perturbations of disease-relevant cell states and tissue environments. Regulatory T cell modulation in atopic dermatitis was predicted to suppress the pro-inflammatory tissue environment, supporting manipulation of these cells as a therapeutic target. In kidney cancer, in silico T cell replacement recapitulated immune checkpoint blockade, while spatially targeted macrophage depletion reverted immunosuppressed T cell states. The corresponding gene programs correlated with survival in large kidney cancer patient cohorts. Together, these findings position MintFlow as a tool for unbiased disease mechanism prediction and in silico perturbation, accelerating translational hypothesis generation and guiding therapeutic strategies.

genomics↗

bioRxiv2024

Global-scale quantification of responses to anthropogenic stressors in six riverine organism groups

Rivers globally are impacted by numerous anthropogenic stressors, including water pollution, habitat degradation, and climate change, which collectively stress biodiversity and ecosystem functioning. This study systematically reviews and analyses published and unpublished data to understand how five aquatic organism groups (bacteria, algae, macrophytes, invertebrates, fish) respond to seven common stressors (salinization, oxygen depletion, fine sediment enrichment, temperature increase, flow modifications and nitrogen or phosphorus enrichment). Using an analytical framework that includes Generalized Linear Models (GLMs) and Robust Bayesian Meta Analysis (RoBMA), we extracted data from 143 relevant datasets out of 29,749 screened articles. Our results reveal a negative relationship between invertebrates and salinity, fine sediment enrichment, and temperature increase, while fish respond positively to increased oxygen levels and temperature. Bacteria and algae show variable responses, with algae positively associated with nitrogen. The findings highlight strong variability in stressor-response relations across organism groups and stressor types, and emphasize the need for more targeted studies on underrepresented groups like macrophytes and microorganisms. This analysis enhances the predictive understanding of stressor impacts on riverine biodiversity, informing future river ecosystem management and restoration efforts.

ecology↗

bioRxiv2024

Large-scale characterization of cell niches in spatial atlases using bio-inspired graph learning

Spatial omics allow us to identify and analyze communities of cells coordinating specific functions within a tissue. While these communities, defined as cell niches, are fundamentally shaped by interactions between spatially neighboring cells, we lack computational frameworks that can leverage spatial omics data to quantitatively characterize niches based on cell interaction events. To address this, we introduce NicheCompass, a graph deep learning method designed based on the principles of cellular communication. NicheCompass not only identifies cell niches, but also learns and informs about the signaling events shaping the identity of these niches. Unlike existing methods, it uniquely characterizes niches by quantifying their activity of spatial gene programs which represent diverse mechanisms of cell-cell communication and transcriptional regulation, thereby uncovering the underlying cellular processes constituting each niche. We showcase a comprehensive workflow encompassing data integration, niche identification, and functional interpretation, and demonstrate that, with its biologically informed design, NicheCompass outperforms existing methods. NicheCompass is broadly applicable to spatial transcriptomics data, which we illustrate by mapping the architecture of diverse tissues during mouse embryonic development, and delineating basal (KRT14) and luminal (KRT8) tumor niches in human breast cancer. We further introduce fine-tuning-based spatial reference mapping, revealing an SPP1+ macrophage-dominated tumor niche in non-small cell lung cancer patients. Additionally, we extend NicheCompass to multimodal spatial profiling of gene expression and chromatin accessibility, identifying and characterizing distinct white matter niches in the mouse brain. Finally, we apply NicheCompass to a whole mouse brain spatial atlas with 8.4 million cells demonstrating its scalability and ability to build foundational, interpretable spatial representations for entire organs. Overall, NicheCompass provides a novel approach to the challenge of identifying and analyzing niches, and suggests a more rigorous niche definition grounded in the quantitative characterization of underlying cellular processes.

bioinformatics↗