Single-cell and spatial transcriptomics

The paper trains a supervised VAE on bulk RNA-seq data to learn a unified tissue-aware representation, achieving high tissue classification accuracy and demonstrating the impact of dataset scale.

This paper uses surprisal information-theoretic analysis on spatial transcriptomic data from human glioblastoma to uncover hidden spatial programs and interactions not detected by standard methods.

The paper shows that immune organization is reflected in coordinated transcriptional changes across cell types, forming a shared low-dimensional state space that persists across tissues and time.

Navigo models mouse embryogenesis to predict cell fates and disease by learning developmental trajectories from scRNA-seq data.

HoloCell is a generative foundation model that integrates epigenomics, transcriptomics, and proteomics data across single-cell multi-omics modalities, demonstrating superior performance in representation learning and cross-modal generation.

FateLimit quantifies when future cell identities become predictable from current molecular states across multiple developmental systems.

OmniCell integrates single-cell and spatial transcriptomics to model tissue context, revealing how gene programs operate in different anatomical environments.

VCBench evaluates single-cell foundation models across seven dimensions, finding baselines often outperform models, with TranscriptFormer showing unique strengths but tradeoffs.

PISTACHIO improves spatial cell-type deconvolution in tumours using proteomics constraints and non-negative matrix factorization.

The paper benchmarks dopaminergic stem cell models and cell therapy candidates using a human brain atlas to identify off-target populations and lineage plasticity.

This study benchmarks 20 cell type annotation methods on spatial transcriptomics data to evaluate their performance in resolving cell types, subtypes, and biological organization across diverse datasets.

ATLAS integrates single-cell RNA-seq and ATAC-seq data to improve trajectory inference by jointly analyzing transcriptomic and chromatin accessibility information.

CellTosg2Sequence integrates biomedical text and signaling graphs into a language model for single-cell analysis, improving performance over existing models.

The paper evaluates deconvolution methods' robustness to incomplete reference matrices and introduces DICEPro to improve performance by adjusting reference signatures.

The paper introduces Suco and Compocyte to classify immune cells from 3,965 patients, identifying new macrophage and monocyte states and T cell memory erosion in metastases.

The paper quantifies seed-induced clustering variability in scRNA-seq and introduces SABER to improve stability and reduce computational cost.

This study evaluates coexpression analysis methods in single-cell and bulk data, finding that xCell provides more replicable and biologically interpretable results than other modes.

Morpho-FM predicts spatial gene expression from H&E images using a pretrained transcriptomic foundation model, achieving strong performance in prostate and kidney cancer datasets.

This paper introduces eVAE, a discrete-latent model for single-cell perturbation prediction, achieving state-of-the-art results on multiple metrics and outperforming existing methods in perturbation ranking.

The paper identifies context-dependent transcriptional states in normal breast tissue that show population-associated differences, with distinct biological programs in specific tissue contexts.

The paper benchmarks gene expression reconstruction from single-cell latent representations, finding autoencoders perform best at low dimensionality and foundation models depend on decoder architecture.

This paper introduces STACAME, a framework for analyzing cross-species spatial transcriptomics data to identify shared and divergent tissue architectures across species.

The paper identifies a widespread transcript leakage issue in spatial transcriptomics and introduces DeLeakage, a Bayesian method that improves data cleaning and cell-type annotation.

This study identifies multicellular spatial molecular programs in meningiomas using multi-omic data and develops a model to infer these programs from histopathology.

Steller improves clonal detection sensitivity in spatial transcriptomics and reveals region-specific spatial clonal architectures in the developing mouse forebrain.

This paper creates a detailed heart atlas using multimodal data to identify cell states and disease-associated genetic variants in specific heart regions.

The paper evaluates whether spatial associations between hypoxia and cytotoxic gene activity in breast cancer sections are consistent across patients, finding they are not.

DOMINO identifies spatial domains in spatial transcriptomic data with improved accuracy and scalability, outperforming existing methods on large datasets.

The paper uses single-nucleus multiomics to study pituitary cell changes during the mouse estrous cycle, finding stage-dependent gene regulatory circuits and validated ETS2-driven gene expression patterns.

This paper introduces xVDM, a method for three-dimensional molecular phenotyping in intact tissues, achieving dense molecular networks and broader transcriptome recovery than traditional approaches.