Jensen–Shannon divergence framework for quantifying gene-centric differences between matched bulk and single-cell RNA-seq breast cancer datasets

Deep learning models rely heavily on extensive training data, but obtaining sufficient real-world data remains a major challenge in clinical fields. To address this, we explore methods for generating realistic synthetic multivariate fall data to supplement limited real-world samples collected from three fall-related datasets: SmartFallMM, UniMib, and K-Fall. We apply three conventional time-series augmentation techniques, a Diffusion-based generative AI method, and a novel approach that extracts fall segments from public video footage of older adults. A key innovation of our work is the exploration of two distinct approaches: video-based pose estimation to extract fall segments from public footage, and Diffusion models to generate synthetic fall signals. Both methods independently enable the creation of highly realistic and diverse synthetic data tailored to specific sensor placements. To our knowledge, these approaches and especially their application in fall detection represent rarely explored directions in this research area. To assess the quality of the synthetic data, we use quantitative metrics, including the Fréchet Inception Distance (FID), Discriminative Score, Predictive Score, Jensen-Shannon Divergence (JSD), and Kolmogorov-Smirnov (KS) test, and visually inspect temporal patterns for structural realism. We observe that Diffusion-based synthesis produces the most realistic and distributionally aligned fall data. To further evaluate the impact of synthetic data, we train a long short-term memory (LSTM) model offline and test it in real time using the SmartFall App. Incorporating Diffusion-based synthetic data improves the offline F1-score by 7-10% and boosts real-time fall detection performance by 24%, confirming its value in enhancing model robustness and applicability in real-world settings.

USP39 is involved in mRNA splicing and stress responses. This study analyzed USP39 expression and its clinical relevance in sepsis, aiming to provide a novel treatment target for sepsis. The RNA-seq datasets were used for analyzing the expression profile and clinical significance of USP39 in both sepsis and control groups. Functional pathway enrichment was performed using GSEA and GSVA with the clusterProfiler package (3.14.3). Immune infiltration was evaluated via single-sample GSEA (ssGSEA). The prognostic value of USP39 was assessed utilizing Cox regression, and its diagnostic performance was determined via receiver operating characteristic (ROC) curve analysis employing the survival (3.5.7) and pROC (1.18.5) packages, respectively. For single-cell RNA-seq (scRNA-seq) data (GSE175453), we performed quality control, normalization, principal component analysis (PCA), batch correction, clustering, and Uniform Manifold Approximation and Projection (UMAP) visualization employing the Seurat 4.4.0 package. THP-1 cells were treated with lipopolysaccharide (LPS) to mimic septic injury. USP39 was overexpressed via pcDNA3.1 transfection. Cellular assays were conducted to measure cell viability, apoptosis, and inflammatory cytokines (IL-1β, IL-6, TNF-α). Our study found that USP39 was significantly downregulated in sepsis patients and was associated with age, diabetes, and Intensive Care Unit (ICU)-acquired infections. Functional enrichment analysis revealed that high USP39 expression was linked to metabolic processes, primary immunodeficiency, and spliceosome, and immune response and inflammation pathway. In addition, USP39 also exhibited a high diagnostic value and was identified as an independent prognostic marker in sepsis. ssGSEA revealed higher infiltration of CD8+ T cells, activated B cells, and CD4+ T cells in the high USP39 expression group, while Th17 and NK cells were more abundant in the low USP39 expression group. According to the results of scRNA-seq analysis, USP39 exhibited significant differential expressions in monocytes and T cells, with low expression observed in sepsis. Furthermore, in vitro experiments confirmed that overexpression of USP39 significantly enhanced cell viability and suppressed apoptosis in LPS-induced THP-1 cells while also inhibiting the expression of inflammatory cytokines. The significant downregulation of USP39 was correlated with both clinical features and immune infiltration in sepsis, highlighting its role in immune regulation and its potential as a prognostic biomarker.

Chronic kidney disease (CKD) remains a global health challenge characterized by high morbidity and mortality, yet its molecular mechanisms remain incompletely defined. Both mitochondrial dysfunction and macrophage polarization have been implicated in CKD pathogenesis, but the precise gene networks and cellular contexts driving these processes are poorly understood. An integrative analysis of bulk RNA sequencing, single-cell transcriptomics, and clinical validation was performed to identify key genes linking mitochondrial regulation and macrophage polarization in CKD. Differentially expressed genes were intersected with mitochondria- and macrophage-related gene sets, refined by machine learning, and assessed through functional enrichment, immune infiltration, and network analyses. Single-cell RNA-seq was applied to resolve cellular heterogeneity and ligand-receptor interactions, while experimental validation was carried out in CKD patient peripheral blood samples and a unilateral ureteral obstruction (UUO) mouse model using qPCR, immunohistochemistry, and Western blotting. A total of METTL17 and SLC27A1 were identified as consistently dysregulated genes across datasets, with METTL17 downregulated and SLC27A1 upregulated. Single-cell analysis localized these alterations primarily to proximal tubular cells (PTCs) and smooth muscle cells (SMCs), where ligand-receptor signaling mediated pathogenic intercellular communication. Immune infiltration analysis revealed selective alterations in naïve B cells, activated NK cells, and γδ T cells, highlighting crosstalk between metabolic and immune pathways. Importantly, both genes showed positive correlations with serum creatinine in CKD patients (e.g., SLC27A1: r = 0.467, P = 0.038), underscoring their clinical relevance. Experimental validation in CKD patient peripheral blood samples (P < 0.01 for both genes) and in a unilateral ureteral obstruction (UUO) mouse model of renal fibrosis confirmed consistent dysregulation: METTL17 was significantly downregulated and SLC27A1 upregulated at both mRNA and protein levels (e.g., immunohistochemistry, P < 0.05). These alterations were spatially localized to proximal tubules and coincided with marked collagen deposition. This study identifies METTL17 and SLC27A1 as key mediators of CKD progression, bridging mitochondrial dysfunction, metabolic reprogramming, and immune imbalance. These findings provide a translational framework for developing biomarker-driven diagnostics and targeted interventions in CKD.

Single-cell RNA-sequencing is an invaluable research tool that allows for the investigation of gene expression in heterogeneous cancer cell populations in ways that bulk RNA-seq cannot. However, normal (i.e., non tumor) cells in cancer samples have the potential to confound the downstream analysis of single-cell RNA-seq data. Existing methods for identifying cancer and normal cells include copy number variation inference, marker-gene expression analysis, and expression-based clustering. This work aims to extend the existing approaches for identifying cancer cells in single-cell RNA-seq samples by incorporating variant calling and the identification of putative driver alterations. We found that putative driver alterations can be detected in single-cell RNA-seq data obtained with full-length transcript technologies and noticed that a subset of cells in tumor samples are enriched for putative driver alterations as compared to normal cells. Furthermore, we show that the number of putative driver alterations and inferred copy number variation are not correlated in all samples. Taken together, our findings suggest that augmenting existing cancer-cell filtering methods with variant calling and analysis can increase the number of tumor cells that can be confidently included in downstream analyses of single-cell full-length transcript RNA-seq datasets.

The increasing use of electronic health records (EHRs) in medical research and AI-driven healthcare necessitates high-fidelity synthetic data that balances patient privacy with statistical and clinical utility. Traditional generative models, such as generative adversarial networks (GANs) and variational autoencoders (VAEs), struggle with mode collapse, limited sample diversity, and difficulties in modelling complex dependencies in high-dimensional tabular data. This study introduces a diffusion model-based approach for generating synthetic EHR data and evaluates its utility in clustering multimorbidity patterns using Dirichlet process mixture models (DPMMs). Denoising diffusion probabilistic models (DDPMs) iteratively refine noise through a structured denoising process, producing diverse, high-fidelity synthetic records. The DPMM framework, a Bayesian nonparametric clustering method, dynamically determines the number of clusters, effectively handling heterogeneous, imbalanced datasets. Model evaluation incorporates statistical similarity measures, feature correlation analysis, privacy risk assessments, and predictive performance metrics. Results demonstrate that DDPM-generated data surpasses GANs and VAEs in fidelity (Jensen–Shannon divergence (JSD) = 0.020, Pearson pairwise correlation (PPC) = 0.94), and privacy preservation (membership inference attack (MIA) Risk = 0.25). DPMM clustering reveals clinically meaningful disease patterns, outperforming traditional clustering models. These findings highlight the potential of diffusion models for privacy-preserving synthetic EHR generation and robust multimorbidity clustering in healthcare analytics.

Accurate disease phenotype prediction plays an important role in the treatment of heterogeneous diseases like cancer in the era of precision medicine. With the advent of high throughput technologies, more comprehensive multi-omics data is now available that can effectively link the genotype to phenotype. However, the interactive relation of multi-omics datasets makes it particularly challenging to incorporate different biological layers to discover the coherent biological signatures and predict phenotypic outcomes. In this study, we introduce omicsGAN, a generative adversarial network model to integrate two omics data and their interaction network. The model captures information from the interaction network as well as the two omics datasets and fuse them to generate synthetic data with better predictive signals. Large-scale experiments on The Cancer Genome Atlas breast cancer, lung cancer and ovarian cancer datasets validate that (i) the model can effectively integrate two omics data (e.g. mRNA and microRNA expression data) and their interaction network (e.g. microRNA-mRNA interaction network). The synthetic omics data generated by the proposed model has a better performance on cancer outcome classification and patients survival prediction compared to original omics datasets. (ii) The integrity of the interaction network plays a vital role in the generation of synthetic data with higher predictive quality. Using a random interaction network does not allow the framework to learn meaningful information from the omics datasets; therefore, results in synthetic data with weaker predictive signals. Source code is available at: https://github.com/CompbioLabUCF/omicsGAN. Supplementary data are available at Bioinformatics online.

Bulk and single-cell RNA-seq (scRNA-seq) data are being used as alternatives to traditional technology in biology and medicine research. These data are used, for example, for the detection of differentially expressed (DE) genes. Several statistical methods have been developed for the classification of bulk and single-cell RNA-seq data. These feature genes are vitally important for the classification of bulk and single-cell RNA-seq data. The majority of genes are not DE and they are thus irrelevant for class distinction. To improve the classification performance and save the computation time, removal of irrelevant genes is necessary. Removal will aid the detection of the important feature genes. Widely used schemes in the literature, such as the BSS/WSS (BW) method, assume that data are normally distributed and may not be suitable for bulk and single-cell RNA-seq data. In this article, a category encoding (CAEN) method is proposed to select feature genes for bulk and single-cell RNA-seq data classification. This novel method encodes categories by employing the rank of sequence samples for each gene in each class. Correlation coefficients are considered for gene and class with the rank of sample and a new rank of category. The highest gene correlation coefficients are considered feature genes, which are the most effective for classifying bulk and single-cell RNA-seq dataset. The sure screening method was also established for rank consistency properties of the proposed CAEN method. Simulation studies show that the classifier using the proposed CAEN method performs better than, or at least as well as, the existing methods in most settings. Existing real datasets were analyzed, with the results demonstrating superior performance of the proposed method over current competitors. The application has been coded into an R package named "CAEN" to facilitate wide use.

Single-cell RNA sequencing has become a valuable tool for investigating cell types in complex tissues, where clustering of cells enables the identification and comparison of cell populations. Although many studies have sought to develop and compare different clustering approaches, a deeper investigation into the properties of the resulting populations is lacking. Specifically, the presence of misclassified cells can influence downstream analyses, highlighting the need to assess subpopulation purity and to detect such cells. We developed DSAVE (Down-SAmpling based Variation Estimation), a method to evaluate the purity of single-cell transcriptome clusters and to identify misclassified cells. The method utilizes down-sampling to eliminate differences in sampling noise and uses a log-likelihood based metric to help identify misclassified cells. In addition, DSAVE estimates the number of cells needed in a population to achieve a stable average gene expression profile within a certain gene expression range. We show that DSAVE can be used to find potentially misclassified cells that are not detectable by similar tools and reveal the cause of their divergence from the other cells, such as differing cell state or cell type. With the growing use of single-cell RNA-seq, we foresee that DSAVE will be an increasingly useful tool for comparing and purifying subpopulations in single-cell RNA-Seq datasets.

MCOIN: a novel heuristic for determining TFBS motif width

treeclimbR is for analyzing hierarchical trees of entities, such as phylogenies or cell types, at different resolutions. It proposes multiple candidates that capture the latent signal and pinpoints branches or leaves that contain features of interest, in a data-driven way. It outperforms currently available methods on synthetic data, and we highlight the approach on various applications, including microbiome and microRNA surveys as well as single-cell cytometry and RNA-seq datasets. With the emergence of various multi-resolution genomic datasets, treeclimbR provides a thorough inspection on entities across resolutions and gives additional flexibility to uncover biological associations.

Identification of the MIF-CD74/CD44 signaling axis as a therapeutic target in pediatric Acute Myeloid Leukemia

Single-cell RNA-seq (scRNAseq) analyses performed at the cellular level aim to understand the cellular landscape of tissue sections, offer insights into rare cell-types, and identify marker genes for annotating distinct cell types. ScRNAseq analyses are widely applied to cancer research to understand tumor heterogeneity, disease progression, and resistance to therapy. Single-cell data processing is a challenging task due to its high-dimensionality, sparsity, and having imbalanced class(cell-type) distributions. An accurate cell-type identification is highly dependent on preprocessing and quality control steps. To address these issues, generative models have been widely used in recent years. Techniques frequently used include Variational Autoencoders (VAE), Generative Adversarial Networks (GANs), Gaussian-based methods, and, more recently, Flow-based (FB) generative models. We developed a Masked Affine Autoregressive transform-embedded FB (MAF-FB) model. Then, to improve MAF-FB further, we incorporated a mixture of experts (MOE) of attention mechanisms on top of it, resulting in our proposed MOE-FB model. We conducted a comparative analysis of fundamental generative models, aiming to serve as a preliminary guidance for developing novel automated scRNAseq data analysis systems. We performed a large-scale analysis by combiningfour datasets derived from pancreatic tissue sections and for further generalizability assessments, we employed Peripheral Blood Mononuclear Cells (PBMC68K and PBMC3K) and Human Cell Atlas Bone Marrow (HCA-BM10K) datasets. We utilized VAE, GAN, Gaussian Copula, and Automated Cell-Type-informed Introspective Variational Autoencoder (ACTIVA), and compared them against our two novel FB models, MAF-FB and MOE-FB for ScRnaseq synthesis. To evaluate the performances of generative models, we used various discrepancy metrics and performed automated cell-type classification tasks. We also identified differentially expressed genes for each cell type, and inferred cell-cell interactions based on ligand-receptor bindings across distinct cell-type pairs. Among the generative models, FB models, especially MOE-FB, consistently outperformed others across all experimental setups in both discrepancy metrics with comparison to the baseline test set and cell-type classification tasks (with an F1-score of 0.90 precision of 0.89 and recall of 0.92 for the integrated pancreatic datasets). MOE-FB produced biologically more relevant synthetic data, and ligand–receptor–based cell–cell interactions inferred from the synthetic cells closely resemble the original data, achieving an RMSE of 0.65 against the corresponding pancreatic test set. These findings highlight the potential and promising use of FB models, especially MOE-FB, in scRNAseq analyses.

Reactive oxygen species (ROS) play a pivotal role in myocardial infarction (MI), contributing to oxidative stress, inflammation, and tissue remodeling. However, ROS-related gene signatures with diagnostic and mechanistic relevance in MI remain insufficiently defined. Transcriptomic data from six MI cohorts were integrated, with GSE66360 as the training set and five datasets as external validation. Batch correction was performed using ComBat. ROS pathway activity was assessed by single-sample gene set enrichment analysis (ssGSEA). Differentially expressed genes (DEGs) shared across datasets were intersected with ROS-related genes to construct an elastic net logistic regression model. Model performance was evaluated using ROC, calibration, and decision curve analysis. SHAP analysis was conducted for interpretability. Upstream transcription factor and miRNA interactions were predicted, and single-gene GSEA was used to explore biological pathways. Immune infiltration and checkpoint expression were analyzed using multiple deconvolution algorithms. Human AC16 cardiomyocytes were used to explore the functions of ADAM9 in MI. To validate our findings in vivo, we established the mice MI model and performed histology, immunostaining, and qRT-PCR to examine the six signature genes. ROS pathway activity was consistently elevated in MI samples across all cohorts. A six-gene signature (MMP9, ADAM9, BST1, TLR4, CLEC7A, CYP1B1) showed strong diagnostic performance. SHAP analysis identified MMP9 as the top contributor. Regulatory network analysis highlighted NFKB1, STAT1, and miR-21-5p as upstream regulators. Functional enrichment revealed an association with inflammatory and immune pathways. Software algorithm predictions from patient blood cell samples showed that the MI sample exhibited increased infiltration of macrophages, dendritic cells, and fibroblasts, along with upregulation of immune and inflammatory genes. Several model genes correlated positively with endothelial cell infiltration. Function studies in human AC16 cardiomyocytes suggested that ADAM9 inhibited cardiomyocyte survival and enhanced oxidative stress. Experimental validation confirmed that all six genes were significantly upregulated at both mRNA and protein levels in infarcted mouse hearts. We identified a ROS-related six-gene diagnostic signature for MI, with strong performance and mechanistic links to immune activation and vascular remodeling. This model may aid early diagnosis and provide insight into redox-immune interplay in MI.

Motivation and Aim: The subset of CD8 + HLA-DR + regulatory T lymphocytes (CD8 + Treg) was first described in 2014 [1].These cells are known to suppress effector T cells with the involvement of checkpoint inhibitor molecules and have some properties of conventional CD4 + Treg [2].Nevertheless, the nature and function of this subset remain poorly understood.Moreover, the study of the properties of this population becomes important in the context of general age-associated changes in the human immune system and higher vulnerability of CD8 + T lymphocytes to these changes.In preliminary experiments, the CD8 + HLA-DR + population was found to segregate into two subpopulations based on surface expression of CD127 (IL7R).Based on this, the CD8 + Treg phenotype was defined as CD3 + CD8 + HLA-DR + CD127 low .In this work, we aimed to determine transcriptional signatures and age-related changes in gene expression within the CD8 + Treg population using publicly available single-cell RNA-seq (scRNAseq) data.Our objectives were as follows: (1) Identify the CD8 + Treg cluster;(2) Determine transcriptional signatures of CD8 + Treg; (3) Evaluate age-related gene expression changes in the CD8 + Treg cell population.Methods and Algorithms: In this study, we analyzed peripheral blood mononuclear cells (PBMCs) scRNA-seq data from 982 donors, aged 19 to 97 y.o., from the OneK1K cohort [3] and 99 healthy donors, aged 22 to 75 y.o., from the SLE study [4].The Scanpy package (version 1.9.8) [5] was used to analyze scRNA-seq data.Initially, data preprocessing steps such as normalization, logarithmization, scaling, dimensionality reduction (by PCA), and batch correction were performed.Subsequently, cells were clustered using the Leiden algorithm.The Celltypist package [6] was utilized to annotate the cell clusters and identify cluster of CD8 + T effector cells (CD8 + Teff).Then, to more accurately represent CD8+Teff cells in reduced dimensionality embedding, the cells in a given cluster were returned logarithmized gene counts and the preprocessing steps from scaling to batch correction were performed repeatedly.Next, re-clustering and further manual annotation were conducted to identify the CD8 + HLA-DR + CD127 low cell subpopulation.The cell populations were pooled to conduct a pseudobulk differential expression analysis using edgeR (version 4.0.16)[7].Functional analysis of the differentially expressed genes was performed using the g:Profiler web tool [8].The described analyses are available as Jupyter-notebooks on a GitHub page:

Aortic dissection (AD) is a life-threatening condition in which the inner layer of the aorta tears. It has been reported that metabolic syndrome (MS) has a close linkage with aortic dissection. However, the inter-relational mechanisms between them were still unclear. This article explored the hub gene signatures and potential molecular mechanisms in AD and MS. We obtained five bulk RNA-seq datasets of AD, one single cell RNA-seq (scRNA-seq) dataset of ascending thoracic aortic aneurysm (ATAA), and one bulk RNA-seq dataset of MS from the gene expression omnibus (GEO) database. Identification of differentially expressed genes (DEGs) and key modules via weighted gene co-expression network analysis (WGCNA), functional enrichment analysis, and machine learning algorithms (Random Forest and LASSO regression) were used to identify hub genes for diagnosing AD with MS. XGBoost further improved the diagnostic performance of the model. The receiver operating characteristic (ROC) and precision-recall (PR) curves were developed to assess the diagnostic value. Then, immune cell infiltration and metabolism-associated pathways analyses were created to investigate immune cell and metabolism-associated pathway dysregulation in AD and MS. Finally, the scRNA-seq dataset was performed to confirm the expression levels of identified hub genes. 406 common DEGs were identified between the merged AD and MS datasets. Functional enrichment analysis revealed these DEGs were enriched for applicable terms of metabolism, cellular processes, organismal systems, and human diseases. Besides, the positively related key modules of AD and MS were mainly enriched in transcription factor binding and inflammatory response. In contrast, the negatively related modules were significantly associated with adaptive immune response and regulation of nuclease activity. Through machine learning, nine genes with common diagnostic effects were found in AD and MS, including MAD2L2, IMP4, PRPF4, CHSY1, SLC20A1, SLC9A1, TIPRL, DPYD, and MAPKAPK2. In the training set, the AUC of the hub gene on RP and RR curves was 1. In the AD verification set, the AUC of the Hub gene on RP and RR curves were 0.946 and 0.955, respectively. In the MS set, the AUC of the Hub gene on RP and RR curves were 0.978 and 0.98, respectively. scRNA-seq analysis revealed that the SLC20A1 was found to be relevant in fatty acid metabolic pathways and expressed in endothelial cells. Our study revealed the common pathogenesis of AD and MS. These common pathways and hub genes might provide new ideas for further mechanism research.