PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “representation learning”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 55 records · Page 3Linked to original sources
PubMed2026

Dissecting spatial patterning and signaling with directional diffusion in spatial multi-omics.

Spatial multi-omics sequencing enables the simultaneous profiling of transcriptomics, proteomics, and epigenomics at a spatial resolution, offering insights into complex tissue organization and molecular regulation. However, the effective integration of multiple omics modalities in a spatial context remains a major challenge. Here, we present SpaDDM, a spatial multi-omics integration framework based on directional diffusion models (DDMs), which supports spatial pattern identification, cross-omics alignment, and inter-and intracellular signaling flow analysis. SpaDDM employs DDM-based graph networks to learn omics-specific representations by jointly incorporating spatial coordinates and molecular measurements within each modality, followed by an attention mechanism to align features across modalities. We benchmarked SpaDDM on diverse spatial multi-omics datasets, including transcriptomics-epigenomics and transcriptomics-proteomics combinations across multiple tissues and species. SpaDDM consistently outperformed existing methods by more accurately deciphering spatial tissue patterns and effectively reducing the boundary noise between spatial regions. Moreover, the learned low-dimensional coembedded representations of individual cells serve as integral mediators for inferring the signaling flows that underlie spatial patterning. Finally, we demonstrated that SpaDDM alignment of complementary information across multi-omics layers facilitates cross-omics translation and significantly improves the prediction of cell state alignments.

Multiomics↗
PubMed2026

Scalable, generalizable and uncertainty-aware integration of spatial multiomics across diverse modalities and platforms with SCIGMA.

Recent advances in spatial omics technologies have enabled simultaneous profiling of transcriptomic, proteomic, epigenomic, metabolomic and imaging data at high spatial resolution, offering unprecedented opportunities to dissect tissue complexity. However, integrating these diverse and large-scale spatial multimodal datasets remains a major computational challenge. We present SCIGMA, a scalable and generalizable deep learning framework for spatial multiomics integration. SCIGMA introduces an uncertainty-aware contrastive learning objective and multiview graph neural networks to preserve modality-specific signals while learning biologically meaningful joint representations. Unlike previous methods, SCIGMA provides spatially resolved uncertainty estimates, interpretably identifying regions of biological or technical heterogeneity. SCIGMA supports integration of up to five modalities, and its modular framework is extensible to future technologies with even more modalities. It also scales to more than 1 million spatial locations, enabling analysis of high-resolution datasets such as Visium HD and Xenium Prime. We evaluated SCIGMA across 19 datasets spanning 8 modalities, 10 tissues and 9 platforms. On benchmarkable datasets, SCIGMA outperformed other methods in spatial domain detection, modality preservation, feature reconstruction and reproducibility. SCIGMA identifies biologically meaningful structures, refined spatial domains and modality-specific regulatory programs, providing a robust, flexible and future-ready solution for scalable spatial multimodal integration.

Multiomics↗
PubMed2025

scMGCL: accurate and efficient integration representation of single-cell multi-omics data.

MOTIVATION: Single-cell multi-omics data integration is essential for understanding cellular states and disease mechanisms, yet integrating heterogeneous data modalities remains a challenge. We present scMGCL, a graph contrastive learning framework for robust integration of single-cell ATAC-seq and RNA-seq data. Our approach leverages self-supervised learning on cell-cell similarity graphs, in which each modality's graph structure serves as an augmentation for the other. This cross-modality contrastive paradigm enables the learning of biologically meaningful, shared representations while preserving modality-specific features. RESULTS: Benchmarking against state-of-the-art methods demonstrates that scMGCL outperforms others in cell-type clustering, label transfer accuracy, and preservation of marker-gene correlations. Additionally, scMGCL significantly improves computational efficiency, reducing runtime and memory usage. The method's effectiveness is further validated through extensive analyses of cell-type similarity and functional consistency, providing a powerful tool for multi-omics data exploration. AVAILABILITY AND IMPLEMENTATION: Code and datasets are released at https://github.com/zlCreator/scMGCL.

Single-Cell Analysis↗
PubMed2026

ChromBERT-tools: a versatile toolkit for context-specific regulatory representations of transcription regulators across different cell types.

SUMMARY: Representations that encode the genome-wide regulatory behavior of transcription regulators provide a foundation for flexible transcription modeling and in silico regulatory analysis. Existing regulator representations are commonly derived from gene co-expression, motif annotations, or static protein features, which capture useful but limited aspects of regulator identity but do not directly model how regulators participate in region-specific regulatory programs across the genome. ChromBERT addresses this gap by learning context-aware regulatory representations from large-scale ChIP-seq data. However, routine bioinformatics applications require lightweight, accessible, and modular tools for generating, adapting, and interpreting these representations in user-defined biological contexts. Here, we present ChromBERT-tools, a user-oriented toolkit built upon ChromBERT that converts its regulatory representation framework into practical workflows for customizable analysis across cellular contexts. ChromBERT-tools provides command-line interfaces and Python APIs organized into three functional layers: representation generation, predictive modeling, and regulatory interpretation. The representation generation layer produces representations of genomic regions and transcription regulators. The predictive modeling layer fine-tunes ChromBERT for genome-wide regulatory activity prediction through classification or regression tasks, with optimized implementation to reduce running time and computational resource requirements. The regulatory interpretation layer supports inference of the context-specific roles of cis-regulatory elements and transcription regulators. These modules can be used independently or integrated into end-to-end workflows, enabling flexible analyses across diverse datasets. ChromBERT-tools lowers the barrier to applying context-specific regulatory representations in routine genomic analyses. AVAILABILITY AND IMPLEMENTATION: ChromBERT-tools is freely available at https://github.com/TongjiZhanglab/ChromBERT-tools, with documentation at https://chrombert-tools.readthedocs.io/en/latest/. A frozen archival snapshot is available on Zenodo under DOI: 10.5281/zenodo.20094206.

Software↗
PubMed2025

Deep learning-based cell-specific gene regulatory networks inferred from single-cell multiome data.

Gene regulatory networks (GRNs) provide a global representation of how genetic/genomic information is transferred in living systems and are a key component in understanding genome regulation. Single-cell multiome data provide unprecedented opportunities to reconstruct GRNs at fine-grained resolution. However, the inference of GRNs is hindered by insufficient single omic profiles due to the characteristic high loss rate of single-cell sequencing data. In this study, we developed scMultiomeGRN, a deep learning framework to infer transcription factor (TF) regulatory networks via unique integration of single-cell genomic (single-cell RNA sequencing) and epigenomic (single-cell ATAC sequencing) data. We create scMultiomeGRN to elucidate these networks by conceptualizing TF network graph structures. Specifically, we build modality-specific neighbor aggregators and cross-modal attention modules to learn latent representations of TFs from single-cell multi-omics. We demonstrate that scMultiomeGRN outperforms state-of-the-art models on multiple benchmark datasets involved in diseases and health. Via scMultiomeGRN, we identified Alzheimer's disease-relevant regulatory network of SPI1 and RUNX1 for microglia. In summary, scMultiomeGRN offers a deep learning framework to identify cell type-specific gene regulatory network from single-cell multiome data.

Deep Learning↗
PubMed2026

Conditional Diffusion Model-Based Method for Annotation of Antibiotic Resistance Gene Properties.

The crisis of bacterial antibiotic resistance, which has led to a decline in the effectiveness of antibiotics originally used to combat bacterial infections, has emerged as an urgent challenge for public health. Antibiotic resistance genes (ARGs) are one of the key reasons for bacteria to develop resistance to antibiotics. Therefore, accurately identifying and annotating the critical properties of ARGs is of great importance for addressing the antibiotic resistance emergency. Although existing deep learning models demonstrate remarkable effectiveness in extracting local features from sequence data, they still face limitations in the capacity to further gain the enriched latent representations within the data. To address the critical challenge of extracting higher-quality representations from ARGs sequence data, we propose a novel ARGs properties annotation method based on the conditional diffusion model which is used to learn latent representations through domain-specific knowledge injection. Specifically, during the conditional information integration phase, we systematically incorporate ARGs' domain knowledge to guide the diffusion process in generating high-quality latent representations. To overcome information redundancy caused by direct concatenation of conditional information and intermediate features, we design a cross-attention mechanism that enables feature fusion between heterogeneous information sources, thereby enhancing further the quality of obtained representations. Experimental results on widely used data sets demonstrate the framework's effectiveness in achieving superior prediction performance compared to existing methods.

Anti-Bacterial Agents↗
PubMed2026

BioNeuralNet: a graph neural network based Multi-Omics network data analysis tool.

SUMMARY: Multi-omics data offer unprecedented insights into complex biological systems, yet their high dimensionality, sparsity, and intricate interactions pose significant analytical challenges. Network-based approaches have advanced multi-omics research by effectively capturing biologically relevant relationships among molecular features (e.g., genes, proteins, metabolites). While these methods are powerful for representing molecular interactions, there remains a need for tools specifically designed to effectively utilize these network representations across diverse downstream analyses. To fulfill this need, we introduce BioNeuralNet, a flexible and modular Python framework tailored for end-to-end network-based multi-omics data analysis. BioNeuralNet leverages Graph Neural Networks (GNNs) to learn biologically meaningful low-dimensional representations from multi-omics networks, converting these complex molecular networks into versatile embeddings. BioNeuralNet supports all major stages of multi-omics network analysis, including several network construction techniques, generation of low-dimensional representations, and a broad range of downstream analytical tasks. Its extensive utilities, including diverse GNN architectures, and compatibility with established Python packages (e.g., scikit-learn, PyTorch, NetworkX), enhance usability and facilitate quick adoption. BioNeuralNet is an open-source, user-friendly, and extensively documented framework designed to support flexible and reproducible multi-omics network analysis in precision medicine. AVAILABILITY AND IMPLEMENTATION: The BioNeuralNet library is available via The Python Package Index (PyPI). Source code, documentation, tutorials, and workflows are hosted at https://bioneuralnet.readthedocs.io. Code archived at https://doi.org/10.5281/zenodo.17503083.

Graph Neural Networks↗
PubMed2025

Integration of single cell multiomics data by deep transfer hypergraph neural network.

Multi-omics characterization of individual cells offers remarkable potential for analyzing the dynamics and relationships of gene regulatory states across millions of cells. How to integrate multimodal data is an open problem, existing integration methods struggle with accuracy and modality-specific biological variation retention. In this paper, we present scHyper (scalable, interpretable machine learning for single cell integration), a low-code and data-efficient deep transfer model designed for integrating paired and unpaired single-cell multimodal data. We benchmark scHyper against datasets from different multimodal data. ScHyper learns a low-dimensional representation and aligns the covariance matrices of the measured modalities, achieving high accuracy even with large scale atlas-level datasets with low memory and computational time across different cell lines, shedding light on regulatory relationships between different types of omics. Altogether, we show that scHyper is a versatile and robust tool for cell-type label transfer and integration from multimodal single-cell datasets.

Single-Cell Analysis↗
PubMed2025

Refining sequence-to-activity models by increasing model resolution.

Decoding the cis-regulatory syntax that controls gene expression is essential for improving our understanding of cell differentiation and disease. To identify regulatory motifs and their regulatory syntax, deep learning based sequence-to-activity (S2A) models learn transcription factor binding motifs and their combinations from DNA sequence by modeling measured chromatin accessibility. Previously, we developed AI-TAC, a S2A model that predicts chromatin accessibility across various immune cell types in multi-task fashion, effectively decoding the regulatory syntax underlying immune cell differentiation. While ATAC-seq is commonly used to measure regional accessibility, it also provides high-resolution profiles, the distribution of Tn5 insertion sites, that offer additional insights into the precise location and strength of TF binding sites. Here we demonstrate that modeling ATAC-seq profiles alongside accessibility consistently improves predictions of differential chromatin accessibility across cell types. Moreover, we also find that multi-task learning across related immune cell types consistently outperforms single-task models. To understand what additional information bpAITAC learns from ATAC-seq profiles, we systematically compare sequence attributions from models trained with and without ATAC-seq profiles. We identify novel motifs with strong effect sizes that emerge only when profile data is included. Our findings suggest that modeling ATAC-seq at base-pair resolution enables the model to learn a more nuanced and sensitive representation of the cis-regulatory syntax driving immune cell-specific chromatin landscapes.

ATAC-seq↗
PubMed2026

CAKR: commutative algebra k-mer representations for genomics.

Despite the availability of various sequence analysis models, comparative genomic analysis remains a challenge in genomics, genetics, and phylogenetics. Commutative algebra, a fundamental tool in algebraic geometry and number theory, has rarely been used in data and biological sciences. In this study, we introduce commutative algebra k-mer representations as a nonlinear algebraic framework for analyzing genomic sequences. This representation bridges commutative algebra, algebraic topology, combinatorics, and machine learning to establish a mathematical framework for comparative genomic analysis. We evaluate its effectiveness on three tasks including genetic variant classification, phylogenetic tree reconstruction, and viral classification, typically requiring alignment-based, alignment-free, and machine-learning approaches, respectively. In this work, we show that commutative algebra k-mer representations outperform five state-of-the-art sequence analysis methods across twelve primary datasets, with two additional supplementary fragment-placement benchmarks, especially in viral classification, and maintain relatively stable predictive accuracy as dataset size increases, underscoring scalability and robustness.

Genomics↗
PubMed2025

MO-GCAN: multi-omics integration based on graph convolutional and attention networks.

MOTIVATION: Cancer subtypes play a critical role in disease progression, prognosis, and treatment, making their detection essential for tailoring precision medicine. Studies have shown that multi-omics integration outperforms single-omics approaches in cancer subtyping tasks. However, due to the high-dimensionality of multi-omics data, many existing studies either fail to capture the correlation between true labels and learned features, or lack sufficient capacity to model complex biological representations. These limitations hinder the full potential of leveraging the rich and complementary information embedded in multi-omics datasets. RESULT: We propose a framework that leverages supervised feature learning and classification based on a graph-based learning approach with attention mechanism for cancer subtyping. More specifically, we train graph convolutional network models on each omics dataset to extract latent representations, which are then concatenated to form a comprehensive multi-omics feature embedding. We further develop sample fusion network based on the omics-specific graphs, incorporating the derived features and feeding them into a graph attention model for subtype classification. This two-stage multi-omics framework is applied to eight cancer types, with performance evaluated in terms of test accuracy, training time, macro-averaged precision, recall, and F-score. Experimental results show that the proposed method outperforms state-of-the-art approaches across various cancer types. Additionally, we provide empirical evidence supporting the hypothesis that retaining a limited number of high-confidence edges and utilizing enriched embeddings from intermediate graph neural network layers can improve predictive performance. AVAILABILITY AND IMPLEMENTATION: Data and the code are available at https://github.com/YD-00/MO-GCAN-Updated.git.

Neoplasms↗
PubMed2025

Tomtom-lite: accelerating Tomtom enables large-scale and real-time motif similarity scoring.

SUMMARY: Pairwise sequence similarity is a core operation in genomic analysis, yet most attention has been given to sequences made up of discrete characters. With the growing prevalence of machine learning, calculating similarities for sequences of continuous representations, e.g. frequency-based position-weight matrices (PWMs) and attribution-based contribution-weight matrices, is taking on newfound importance. Tomtom has previously been proposed as an algorithm for identifying pairs of PWMs whose similarity is statistically significant, but the implementation remains inefficient for both real-time and large-scale analysis. Accordingly, we have re-implemented Tomtom as a numba-accelerated Python function that is natively multi-threaded, avoids cache misses, more efficiently caches intermediate values, and uses approximations at compute bottlenecks. Here, we provide a detailed description of the original Tomtom method and present results demonstrating that our re-implementation can achieve over a 1000-fold speedup compared with the original tool on reasonable tasks. AVAILABILITY AND IMPLEMENTATION: Our implementation of Tomtom is freely available as a Python package at https://github.com/jmschrei/memesuite-lite, which can be downloaded via pip install memelite or at https://zenodo.org/records/17008952.

Software↗
PubMed2026

A full review of online education resources available on antifungal stewardship.

BACKGROUND AND OBJECTIVES: Antifungal resistance represents an increasing global threat, driven by the rising burden of fungal disease. Antifungal stewardship (AFS) is a critical component of broader antimicrobial resistance (AMR) efforts, but education in this area remains less established than antibacterial stewardship initiatives. The scope and characteristics of the current landscape of online AFS resources have not yet been systematically described. To identify and evaluate online educational resources focused on fungal disease management and AFS, and assess their accessibility, format, educational design and implementation focus. METHODS: A structured search of internet search engines, distribution platforms and organizational websites was conducted to identify English-language web-based resources related to fungal disease management and stewardship. Resources were evaluated using predefined criteria including access model, format, length, educational design, interactivity and AFS content. An overall educational value score (1-10) was assigned. RESULTS: Twenty-three educational resources were identified. Most were delivered as online unfacilitated courses (11, 48%) and were short (<4&#x2005;h) (12, 52%). Most focused on guidelines and syndromic management (18, 78%) and targeted doctors and/or nurses/midwives (22, 96%). Limited interactivity was reported in nine (39%) courses. Five courses (22%) had either a substantial or comprehensive focus on AFS. CONCLUSIONS: Online AFS educational resources are available and support awareness and knowledge development. However, they remain relatively few in number. Greater emphasis on implementation-focused learning, behaviour change components and broader global representation may enhance their impact.

Journal Article↗
PubMed1977

Nonverbal visual short-term memory as a function of age and dimensionality in learning-disabled children.

A serial recognition task was used to compare performance of 2 learning disability age groups with 2- and 3-dimensional representations of nonlabeled 8-point random shapes. Age-related increases in short-term memory (STM) performance for both dimensions were found. No significant differences were found between 2- or 3-dimensional stimuli. Contrary to reports of STM performance with normal children, learning-disabled children showed no primacy effect for the 2-dimensional treatment, and second choices were not consistently correct when the first choice was incorrect, These findings were interpreted according to Flavell's notions of mediational inefficiencies.

Age Factors↗
PubMed1992

Cognitive mechanisms of face processing.

Evidence from natural and induced errors of face recognition, from the effects of different cues on resolving errors, and from the latencies to make different decisions about seen faces, all suggest that familiar face recognition involves a fixed, invariant sequence of stages. To recognize a familiar face, a perceptual description of a seen face must first activate a long-standing representation of the appearance of the face of the familiar person. 'Semantic' knowledge about such things as the person's occupation and personality are accessed next, followed, in the final stage, by the name. Certain factors affect the ease of familiar face recognition. Faces seen in the recent past are recognized more readily (repetition priming), as are distinctive faces, and faces preceded by those of related individuals (associative priming). Our knowledge of these phenomena is reviewed for the light it can shed upon the mechanisms of face recognition. Four aspects of face recognition--graded similarity effects and part-to-whole completion in repetition priming, prototype extraction with simultaneous retention of information about individual exemplars, and distinctiveness effects in classification and identification--are proposed as being compatible with distributed memory accounts of cognitive representations.

Association Learning↗
PubMed2026

Leveraging Interradiomic Feature Relationships for Enhanced Prediction of Distant Metastasis and Characterization of Heterogeneity in Head and Neck Cancer.

PURPOSE: Distant metastasis remains a major cause of treatment failure in head and neck (HN) cancer, highlighting the need for more accurate early risk stratification. This study developed and validated a deep radiomics framework to characterize tumor heterogeneity from pretreatment computed tomography (CT) images and improve prediction of distant metastasis-free survival (DMFS). METHODS AND MATERIALS: This multicenter study included 3421 patients with HN cancer from 4 cohorts across 12 institutions. Radiomics features were extracted from primary tumors and transformed into OmicsMaps, a structured representation that spatially organizes interfeature relationships to facilitate learning of complex prognostic patterns. A convolutional neural network was trained to derive prognostic signatures, which were integrated with key clinical variables to construct an OmicsMap-clinical fusion model for patient risk stratification. Model performance was assessed using the concordance index (C-index) and time-dependent area under the receiver operating characteristic curve (AUC) in the CT Images from Large Head and Neck Cohort (RADCURE), HEAD-NECK-RADIOMICS-HN1 (HN1), and Head-Neck-Positron Emission Tomography-Computed Tomography (HN-PET-CT) cohorts. Radiogenomic analyses using RNA-seq data were conducted in the Cancer Genome Atlas Head-Neck Squamous Cell Carcinoma (TCGA-HNSC) cohort to investigate biological characteristics associated with the imaging-defined risk groups. RESULTS: The OmicsMap achieved C-index values of 0.742, 0.768, and 0.671 in the RADCURE, HN1, and HN-PET-CT cohorts, outperforming the conventional radiomics approach by 5.40%-6.37%. Incorporating clinical variables further improved generalizability, yielding a C-index of 0.864 (HN1) and 0.730 (HN-PET-CT), with time-dependent AUC of 0.727-0.895. The fusion model consistently stratified patients into distinct high- and low-risk groups for both DMFS and overall survival across cohorts (P <.01). Radiogenomic analyses revealed enrichment of immune-related pathways in the low-risk group, whereas the high-risk group exhibited a more aggressive phenotype enriched for proliferation, hypoxia, and epithelial-mesenchymal transition pathways, along with a fibrosis-prone tumor microenvironment characterized by extracellular matrix remodeling. CONCLUSIONS: Modeling interradiomic feature relationships using the OmicsMap representation substantially improves CT-based prediction of DMFS and characterization of tumor heterogeneity in HN cancer, supporting precision risk stratification in clinical oncology.

Journal Article↗
PubMed2026

A transcription factor regulatory atlas for activity inference and perturbation prediction.

Inferring transcription factor (TF) activity from transcriptomes and predicting transcriptome-wide responses to TF perturbations remain challenging, in part because available TF-mRNA resources often face a trade-off between precision and coverage and typically lack signed regulatory information. Here, we present TFActProfiler, a TF-mRNA resource and computational framework that learns signed, quantitative TF-mRNA regulatory coefficients by integrating heterogeneous prior evidence (ChIP-based, motif-based, and curated TF-mRNA annotations) with large-scale bulk and single-cell RNA-seq atlases. TFActProfiler contains 2&#x2009;606&#x2009;176 signed TF-mRNA interactions and improves TF activity inference in TF knockdown benchmarks relative to widely used regulon resources while retaining broad TF and target coverage. In addition, because the same learned regulatory coefficients can be used to model downstream transcriptional effects, TFActProfiler enables prediction of transcriptome-wide gene expression responses to TF knockdown without training on task-matched perturbation data. When perturbation datasets are available, TFActProfiler can be further refined to achieve performance comparable to state-of-the-art machine-learning baselines. By providing a direction-aware representation of TF-mRNA regulation for both activity inference and perturbation-response modeling, TFActProfiler supports systematic dissection of gene regulatory programs across diverse cellular contexts.

Transcription Factors↗