SEARCH · Search PubMed

Results for “Multiome”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

SourceResearch connectionFrom yearThrough year

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

Multiomic Integration Reveals Novel miRNA-mRNA-Protein Expression Profile in the Aged Female Retina.

PURPOSE: Aging is a leading risk factor for retinal degeneration. MicroRNAs (miRNAs) regulate posttranscriptional gene suppressors and influence inflammation and oxidative stress, two processes disrupted during retinal aging. This study aimed to identify age-related miRNA-mRNA-protein associations between young and older retinas and uncover dysregulated pathways that may contribute to retinal degeneration. METHODS: Retinal function was assessed using electroretinography (ERG), and microgliosis was quantified by microglial immunohistochemistry (IHC). A multiomics approach was used to examine molecular changes in older (30-month-old) female C57BL/6J mouse retinas and compared with young female (3-month-old) controls. Illumina sequencing profiled short miRNAs (20 bp) and bulk mRNAs (150 bp), while total proteomics via mass spectrometry assessed protein expression. Bioinformatic analyses included targetome analysis (miRNet), pathway enrichment (Gene Ontology), and clustering to identify age-associated molecular targets and pathways. RESULTS: Retinas from older mice displayed neuronal dysfunction and increased microgliosis. Sequencing revealed significant dysregulation of miRNAs linked to immune and inflammatory pathways, supported by enrichment of their predicted mRNA targets. In the older mice, mRNA expression showed broad inflammatory activation, though only 14% of dysregulated mRNAs overlapped with predicted miRNA targets. Proteomic profiling revealed a disconnect between RNA and protein expression, yet all omics layers showed enrichment in inflammatory pathways. Integrated analysis identified associations involving several gene regulatory networks in the older retina. CONCLUSIONS: This study demonstrates that at an advanced age, miRNA expression and their predicted downstream regulatory networks are dysregulated, highlighting potential molecular mechanisms underlying age-related retinal degeneration.

Animals↗

PubMed2026

Harnessing probiotics to combat nonylphenol toxicity: a multiomics approach of gut microbiome remodelling in Silurus meridionalis.

BACKGROUND: As a ubiquitous environmental endocrine disruptor, nonylphenol (NP) threatens aquatic organisms, driving the need for sustainable mitigation strategies. While probiotics represent promising eco-friendly supplements, their molecular mechanisms against NP toxicity remain unclear. In this study, S. meridionalis received 7-week of probiotic (Bacillus subtilis and Lactobacillus acidophilus) pretreatment followed by 15 days of NP exposure. Integrated metagenomics, transcriptomics, and metabolomics analyses, with Reverse transcription quantitative real-time PCR (RT‒qPCR) and Enzyme-linked immunosorbent assay (ELISA) validation, were performed to elucidate microbial, genetic and metabolic responses. Growth performance, including the specific growth rate (SGR) and weight gain rate (WGR), was concurrently assessed. RESULTS: NP exposure significantly suppressed WGR and SGR, and induced gut microbiota dysbiosis alongside and lipid metabolism disorders in S. meridionalis. Probiotic pretreatment effectively reversed these toxic effects and restored the inhibited WGR and SGR. Multiomics integration revealed that the protective effects of probiotics were mediated by a coherent "microbe-host" co-metabolism network across 3 progressive layers: (1) Microbial Remodelling: in which beneficial taxa (e.g., Bacteroides eggerthii and Cetobacterium sp.) were enriched, and the functional capacity for short-chain fatty acid (SCFA) synthesis and ethanolamine metabolism was enhanced; (2) Host Gene Regulation: in which key lipid metabolism genes (ek1, cept1, ept1, mogat2, and abcg2a) were upregulated, and lipase activity was restored; and (3) Metabolic Pathway Activation and Physiological Repair: in which the activity of the NP-suppressed Kennedy pathway was reactivated, thereby promoting phosphatidylethanolamine (PE) and phosphatidylcholine (PC) synthesis and ultimately restoring gut barrier function. These results were further were corroborated by RT‒qPCR and ELISA. CONCLUSION: This study systematically elucidated that probiotics alleviated NP toxicity by remodelling a "microbiota-host Kennedy pathway gene-metabolite (PE and PC)-growth performance" regulatory network. The key mechanism is the beneficial microbiota activating the host Kennedy pathway and restoring gut phospholipid homeostasis and barrier function. These findings provide a theoretical basis for developing targeted, lipid metabolism focused probiotic feed additives for use in sustainable aquaculture.

Probiotics↗

PubMed2026

Multiomics analysis reveals that senescent CXCL16+ macrophages promote lung adenocarcinoma progression through TGF-β signalling.

BACKGROUND: Lung adenocarcinoma (LUAD) is the most common histological subtype of lung cancer and remains a leading cause of cancer-related mortality worldwide. Although, immunotherapy has become a cornerstone of first-line treatment, only 20-30% of patients achieve a durable clinical benefit, largely because of the complexity and heterogeneity of the tumour immune microenvironment. Emerging evidence indicates that cellular senescence, particularly within immune cells, contributes to tumour progression by impairing antitumour immunity; however, its mechanistic role in LUAD remains incompletely understood. METHODS: We performed an integrative multiomics analysis incorporating genome-wide association studies (GWASs), bulk RNA sequencing, single-cell RNA sequencing, and spatial transcriptomics to characterize immune heterogeneity in LUAD. Cellular senescence was validated by performing staining for senescence-associated β-galactosidase and the canonical markers p16 and p21. SHAP analysis was applied to evaluate the contribution of CXCL16+ macrophages. Functional roles were assessed using coculture assays, in vitro and in vivo tumour models, orthotopic tumour implantation, and multiplex immunofluorescence staining of clinical specimens. RESULTS: A summary data-based on Mendelian randomization analysis integrating GWAS and TCGA data identified CXCL16 as a senescence-associated gene that is causally linked to the LUAD risk. Single-cell RNA sequencing revealed that CXCL16 is predominantly expressed in macrophages, and the pseudotime analysis together with β-galactosidase staining confirmed its association with macrophage senescence. Spatial transcriptomics and immunofluorescence staining showed the marked enrichment of CXCL16+ macrophages in LUAD tissues. The cell-cell communication analysis further revealed a strong association between the number of CXCL16+ macrophages and the activation of the TGF-β signalling pathway within the tumour microenvironment. Functionally, CXCL16+ macrophages promoted LUAD progression via TGF-β signalling, as validated in vitro and in subcutaneous and orthotopic tumour models. Molecular dynamics simulations additionally suggested that LUAD patients with high levels of CXCL16+ macrophage infiltration may exhibit increased sensitivity to bosutinib. CONCLUSIONS: CXCL16 promotes macrophage senescence, and senescent CXCL16+ macrophages drive LUAD progression through TGF-β signalling. These findings identify CXCL16+ macrophages as a biologically and therapeutically relevant immune cell population, highlighting a potential target for precision intervention in LUAD.

Humans↗

PubMed2026

Single nucleus multiomics reveals an early inflammatory response to high-fat diet in mouse islets.

In periods of sustained hyper-nutrition, pancreatic β-cells undergo functional compensation through transcriptional upregulation of gene programs driving insulin secretion. This adaptation is essential for maintaining systemic glucose homeostasis and metabolic health. Using single nuclei multiomics, we have mapped the early transcriptional adaptive mechanisms in murine islets of Langerhans exposed to high-fat diet (HFD) for 1 and 3 wk. We show that β-cells exhibit the largest transcriptional response to HFD, characterized by early activation of pro-inflammatory eRegulons and down-regulation of β-cell identity genes, particularly in a distinct subset of β-cells. These observations extend to humans, where the prevalence of an β-cells with a high inflammatory signature is increased in diabetes. Collectively, these observations point to cellular crosstalk through pro-inflammatory signaling as a central and early driver of β-cell dysfunction that limits the compensatory capacity of β-cells, which is closely linked to the development of diabetes.

Animals↗

PubMed2026

VINE-seq and MultiVINE-seq for single-nucleus and multiome profiling of the brain vasculature.

The human cerebrovasculature is a critical yet historically understudied component of neurological health. Dysfunction of the diverse endothelial, mural, and perivascular cells that comprise cerebral vessels is central to diseases ranging from stroke to Alzheimer's disease. However, characterizing these cell populations at a molecular level has proven exceptionally challenging. Encased within a robust basement membrane, vascular cells resist standard dissociation methods, leading to their systematic depletion and underrepresentation in existing single-nucleus genomic atlases. This has created a major blind spot in neuroscience. To overcome this barrier, we developed vessel isolation and nucleus extraction for sequencing (VINE-seq) and its advanced iteration, MultiVINE-seq. The protocol provides a robust, reproducible workflow for the enrichment and high-resolution profiling of vascular, perivascular, and immune cells from fresh or frozen human and mouse brain tissue. First, intact vessels (predominantly capillaries and small arterioles/venules, 100 µm in diameter) are isolated from homogenized brain tissue via dextran-based density-gradient centrifugation, separating the vascular pellet from myelin and the parenchymal fraction. Second, the collected vessels are rigorously washed over a cell strainer to remove trapped contaminants. A critical innovation lies in the third stage: the optimized extraction of nuclei from purified vessels using enzymatic digestion. After extraction, the protocol uses fluorescence-activated cell sorting (FACS) to ensure collection of high-purity nuclei suitable for widely used droplet-based sequencing platforms (e.g., 10x Genomics single cell 3' or multiome). This protocol requires 4-5 h to complete and can be carried out by researchers with single-cell and flow cytometry training.

Journal Article↗

PubMed2025

Multiomic clocks to predict phenotypic age in mice.

Biological age refers to a person's overall health in aging, as distinct from their chronological age. Diverse measures of biological age, referred to as "clocks," have been developed in recent years and enable risk assessments and an estimation of the efficacy of longevity interventions in animals and humans. Although most clocks are trained to predict chronological age, clocks have been developed to predict more complex composite biological age outcomes, at least in humans. These composite outcomes can be made up of a combination of phenotypic data, chronological age, and disease or mortality risk. Here, we develop the first such composite biological age measure for mice: the mouse phenotypic age model (Mouse PhenoAge). This outcome is based on frailty measures, complete blood counts, and mortality risk in a longitudinally assessed cohort of male and female C57BL/6 mice. We then develop clocks to predict Mouse PhenoAge, based on multiomic models using metabolomic and DNA methylation data. Our models accurately predict Mouse PhenoAge, and residuals of the models are associated with remaining lifespan, even for mice of the same chronological age. These methods offer novel ways to accurately predict mortality in laboratory mice, thus reducing the need for lengthy and costly survival studies.

Animals↗

PubMed2025

Integrating explainable AI with multiomics systems biology and EHR data mining for personalized drug repurposing in Alzheimer's disease.

Alzheimer's disease (AD) is characterized by region- and patient-specific molecular heterogeneity, which hinders therapeutic design. In this study, we introduce PRISM-ML (PRecision-medicine using Interpretable Systems and Multiomics with Machine Learning), an open-source integrated analysis pipeline that combines interpretable machine learning with systems biology and electronic health record (EHR) data mining to elucidate the molecular diversity of AD and predict promising drug repurposing opportunities. First, we integrated and harmonized transcriptomic (bulk RNA-seq) and genomic (genome-wide association study) data from 2105 brain samples, each with matched data from the same individual (1363 AD patients, 742 controls; nine tissues), sourced from three independent studies. Random forest classifiers with SHapley Additive exPlanations (SHAP) identified patient-specific biomarkers; unsupervised clustering resolved 36 molecularly distinct "subtissues" (clusters of samples); and gene-gene co-expression networks prioritized 262 high-centrality bottleneck genes as putative regulators of dysregulated pathways. Next, knowledge graph-based drug repurposing predicted six FDA-approved drugs that simultaneously target multiple bottleneck genes and multiple AD-relevant pathways. Notably, in a large U.S. de-identified insurance-claims database (n = 364733), exposure to promethazine, one of the candidate drugs, was associated with a 57-62 % lower incidence of AD versus an active antihistamine comparator (adjusted hazard ratio 0.38; inverse-probability weighted 0.43; both p < 0.001), providing real-world support for its repurposing potential. In summary, PRISM-ML, as an explainable multi-omics analysis pipeline, is readily transferable to other complex diseases, advancing precision medicine.

Computational Biology↗

PubMed2025

Integrated Multiomics Analysis of Microsatellite Instability-High Colorectal Cancer Identifies a Subtype With Poor Outcome.

Up to 50% of patients with metastatic microsatellite instability-high (MSI-H) colorectal cancer (CRC) are resistant to immunotherapy and experience progression or recurrence after treatment. We integrated the genomic, epigenomic, transcriptomic, and proteomic data for 99 patients in a Chinese MSI-H CRC cohort. Proteomic profiling of primary tumors clearly classified MSI-H tumors into 2 subtypes. We found that the 2 subtypes have different mutational signatures, enriched pathways, gene fusion networks, and clinical outcomes. Notably, NCAM1 could serve as a potential biomarker for checkpoint inhibitor response in MSI-H CRC. Thus, there is an urgent need to stratify the MSI-H group into different subtypes and adopt more targeted therapies to prolong patient survival.

Humans↗

PubMed2026

Integrated Clinicopathologic and Multiomic Profiling Reveals MEIS1-Rearranged Sarcoma as a Distinct Entity With 2 Prognostic Subgroups.

Sarcomas with MEIS1 fusions represent a rare, recently recognized group of mesenchymal neoplasms with a predilection for genitourinary and gynecologic sites. A subset exhibits skeletal muscle differentiation resembling spindle cell rhabdomyosarcoma. Existing literature is limited to case reports and small series, with scant comprehensive clinicopathologic, molecular, and outcome data. In this study, we analyzed a multi-institutional cohort of 20 MEIS1-rearranged sarcomas using integrated clinicopathologic review, genomic profiling, and DNA methylation analysis. The tumors occurred in 17 females and 3 males (median age, 41 years; range, 6-58 years), arising mainly in the uterus/vagina (n = 12), vulva/perineum (n = 4), bone (n = 2), and kidney (n = 2), with a median size of 9 cm (range, 2.5-20 cm). Histology showed mostly bland spindle cells in fascicles/storiform patterns, alternating cellularity, fibromyxoid stroma, prominent vascularity, and adipose metaplasia (45%). A subset of cases featured high-grade morphology with epithelioid cells and increased mitotic activity. Skeletal muscle markers were variably positive in 9 cases. Fusions involved MEIS1 with NCOA2 (16/20), NCOA1 (3/20), or FOXO1 (1/20). Recurrent additional genomic alterations included CTNNB1 mutations (31.6%) and MDM2 amplification (15%). DNA methylation profiling showed that MEIS1-rearranged sarcomas formed a unifying cluster comprising 2 subgroups, regardless of rhabdomyosarcomatous phenotype, clearly separated from other mesenchymal neoplasms, including various rhabdomyosarcoma subtypes and uterine sarcomas. The 2 DNA methylation (Meth) subgroups correlated with differences in genome-wide copy number variation (CNV) status (Meth-CNV high vs Meth-CNV low), with Meth-CNV high tumors characterized by high mitotic rate, frequent tumor necrosis, recurrent co-occurring CTNNB1 and MDM2 alterations, and recurrent chromosomal arm-level changes. Most importantly, this subgroup exhibited significantly worse overall survival (P = .027) and disease-free survival (median, 5 vs 99 months; P = .017). This study establishes MEIS1-rearranged sarcoma as a distinct entity with generally indolent but potentially aggressive behavior. The 2 methylation/CNV subgroups provide potential utility for prognostic stratification and highlight actionable molecular targets in high-risk cases.

Humans↗

PubMed2025

From genetic causality to druggable targets: A multiomics framework identifies ZSCAN16 in gout pathogenesis.

ObjectiveGout is a prevalent form of inflammatory arthritis in which many patients respond suboptimally to current therapies. Drug development is hampered by a lack of genetically validated targets, leading to high clinical trial attrition. This study aimed to systematically identify and prioritize novel, druggable targets for gout via a multilayered genetic and functional genomics approach.MethodsWe performed two-sample Mendelian randomization (MR) using cis-expression quantitative trait locus (cis-eQTL) data and dual independent gout genome-wide association study (GWAS) cohorts (openGWAS and FinnGen). The candidate genes were subjected to a rigorous validation pipeline including Bayesian colocalization, phenome-wide association studies (PheWASs) to assess pleiotropy and on-target safety, and single-cell RNA sequencing (scRNA-seq) to delineate the cellular context. Molecular docking was used to evaluate the structural druggability of prioritized targets.ResultsMR analysis revealed 15 genes causally associated with gout. Colocalization analysis (PPH4 > 0.8) prioritized two targets: ZSCAN16 (risk-increasing, OR = 1.04, 95% CI [1.02-1.06]) and TRIM10 (protective, OR = 0.96, 95% CI [0.94-0.98]). Crucially, PheWAS revealed that ZSCAN16 is highly specific to gout, whereas TRIM10 exhibited extensive pleiotropy with hematological and cardiometabolic traits, indicating significant safety risks. Single-cell analysis provided orthogonal validation, demonstrating flare-specific upregulation of ZSCAN16 in cytotoxic T/NK cells. Molecular docking confirmed ZSCAN16 as a structurally druggable target, showing high-affinity binding with known compounds (e.g. digoxin, binding energy = -9.6 kcal/mol).ConclusionsOur study identifies ZSCAN16 as a high-potential, druggable therapeutic target for gout, highlighting its genetic influence on specific immune cell activities during acute flares. Conversely, TRIM10 was deprioritized owing to substantial pleiotropic liabilities and poor chemical tractability. These findings suggest that ZSCAN16 could play a crucial role in the pathogenesis of gout and may provide a valuable lead for future drug discovery efforts.

Humans↗

PubMed2025

Multiomics Reveal Associations Between CpG Methylation, Histone Modifications and Transcription in a Species That has Lost DNMT3, the Colorado Potato Beetle.

Insects display exceptional phenotypic plasticity, which can be mediated by epigenetic modifications, including CpG methylation and histone modifications. In vertebrates, both are interlinked and CpG methylation is associated with gene repression. However, little is known about these regulatory systems in invertebrates, where CpG methylation is mainly restricted to gene bodies of transcriptionally active genes. A widely conserved mechanism involves the co-transcriptional deposition of H3K36 trimethylation and the targeted methylation of unmethylated CpGs by the de novo DNA methyltransferase DNMT3. However, DNMT3 has been lost multiple times in invertebrate lineages raising the question of how the links between CpG methylation, histone modifications and gene expression are affected by its loss. Here, we report the epigenetic landscape of Leptinotarsa decemlineata, a beetle species that has lost DNMT3 but retained CpG methylation. We combine RNA-seq, enzymatic methyl-seq and CUT&Tag to study gene expression, CpG methylation and patterns of H3K36me3 and H3K27ac histone modifications on a genome-wide scale. Despite the loss of DNMT3, H3K36me3 mirrors CpG methylation patterns. Together, they give rise to signature profiles for expressed and not expressed genes. H3K27ac patterns show a prominent peak at the transcription start site that is predictive of expressed genes irrespective of their methylation status. Our study provides new insights into the evolutionary flexibility of epigenetic modification systems that urge caution when generalizing across species.

Animals↗

PubMed2026

Multiomic insights into fungal polylactic acid degradation: Metabolic adaptation and hydrolytic mechanisms of Sporobolomyces pararoseus.

Polylactic acid (PLA), a biodegradable polyester from renewable resources, is a sustainable alternative to petrochemical plastics. However, its environmental degradation is inefficient naturally, requiring specific microbial activities. While bacterial PLA-degrading mechanisms are well documented, fungal degrading systems-particularly their molecular mechanisms-are underexplored.We isolated Sporobolomyces pararoseus ZRQ01 from the gut microbiota of PLA-fed mealworms. This fungal strain noticeably degraded PLA in PLA-containing medium supplemented with 2% glucose. Biodegradation assays revealed 22.8% loss of the PLA film weight after 35 days of incubation, and scanning electron microscopy confirmed extensive surface erosion and pore formation. Integrated transcriptomic and proteomic analyses, together with the reference genome of S. pararoseus ZRQ01, revealed that S. pararoseus ZRQ01 upregulates hydrolytic enzymes at both transcript and protein levels to cleave PLA into lactic acid. After lactic acid is transferred into S. pararoseus ZRQ01 cells by monocarboxylate transporters with increased abundance, it is assimilated by pathways of pyruvate metabolism and the TCA cycle with increased protein abundance. Intriguingly, upregulation of genes in autophagy-related and MAPK signaling pathways underscores an adaptive stress response potentially supporting cellular homeostasis and degradation-related gene expression. Our results highlight S. pararoseus ZRQ01's metabolic potential for bioremediation and offer insights into fungal bioplastic degradation pathways.

Polyesters↗

PubMed2026

Multiomics profiling of plasma reveals lipid-immune dysregulation and exosome remodeling in mpox and mpox-HIV co-infection.

BACKGROUND: Monkeypox virus (MPXV) infects diverse human cell types, and human immunodeficiency virus (HIV) co-infection is common. The immunometabolic consequences of MPXV infection, and how it may be altered by HIV, remain poorly defined. METHODS: We performed quantitative plasma lipidomics and precise metabolomics in a discovery cohort (n = 81) comprising MPXV-monoinfected (MPLWOH), MPXV-HIV-coinfected (MPLWH), and HIV-monoinfected (PLWH) patients and healthy controls, integrating exosome proteomics, cytokine profiling, and transcriptomics of exosome-treated HepG2 and A549 cells for functional interpretation. An independent validation cohort (n = 65) was used to assess cross-cohort reproducibility. FINDINGS: MPXV infection induced broad lipid remodeling, with elevations in phosphatidylserine (PS) and phosphatidylethanolamine (PE) and reductions in phosphatidylcholine (PC), lysophospholipids, cholesteryl ester (CE), and exosomal lecithin-cholesterol acyltransferase (LCAT) and lipoprotein lipase (LPL). These lipid alterations were correlated with tissue injury markers and inflammatory cytokines. The MPLWH group exhibited more severe metabolic disruption, including marked sulfatide (SL) depletion, lower cholesterol and high-density lipoprotein cholesterol (HDL-c), and extensive rewiring of lipid-cytokine associations. SL depletion in MPLWH correlated with abundances of COPI-mediated retrograde trafficking proteins in exosomes. Transcriptomic profiling of exosome-treated cells provided functional validation: MPLWOH exosomes induced lipid metabolism and repair-associated epithelial programs, while MPLWH exosomes drove phospholipid remodeling and acute inflammatory and mucosal barrier-stress responses. CONCLUSIONS: MPXV infection reprograms host lipid metabolism and exosome composition, with HIV co-infection amplifying inflammatory, metabolic, and trafficking disruptions. These convergent multi-omics signatures link systemic lipid dysregulation to exosome-mediated immunomodulation and identify potential targets for host-directed interventions. FUNDING: This study was funded by the Major Project of Guangzhou National Laboratory.

Adult↗

PubMed2026

Multiancestry genome-wide association and multiomics analyses elucidate spatiocellular features of multiple sclerosis genetics.

Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system characterized by demyelination disseminated in space and time. Here we performed a genome-wide association study (GWAS) using 688 MS cases and 205,199 controls from the Japanese population and identified significant associations in the major histocompatibility complex region and a population-specific risk variant in 11q24. Through cross-population GWAS meta-analyses using a total of 29,374 cases and 1,843,563 controls from 4 ancestral populations, we identified 22 novel susceptibility loci. Integration of GWAS and single-cell and single-nucleus RNA sequencing of peripheral blood mononuclear cells and subcortical lesions from patients with MS revealed enrichment of genetic risk factors for MS in CD4+ T helper cell lineage and regulatory T cells, as well as in endothelial cells. Furthermore, spatial transcriptomics of subcortical lesions demonstrated spatial and temporal heterogeneity in associations with MS genetic risk. Our study demonstrates the value of investigation of spatiocellular features of disease genetics across diverse populations and omics modalities.

Humans↗

PubMed2025

Temporal multiomics gene expression data of human embryonic stem cell-derived cardiomyocyte differentiation.

Human embryonic stem cells (hESCs) serve as a valuable in vitro model for studying early human developmental processes due to their ability to differentiate into all three germ layers. Here, we present a comprehensive multi-omics dataset generated by differentiating hESCs into cardiomyocytes via the mesodermal lineage, collecting samples at 10 distinct time points. We measured mRNA levels by mRNA sequencing (mRNA-seq), translation levels by ribosome profiling (Ribo-seq), and protein levels by quantitative mass spectrometry-based proteomics. Technical validation confirmed high quality and reproducibility across all datasets, with strong correlations between replicates. This extensive dataset provides critical insights into the complex regulatory mechanisms of cardiomyocyte differentiation and serves as a valuable resource for the research community, aiding in the exploration of mammalian development and gene regulation.

Humans↗

PubMed2025

Transfer learning with multiomics integration and deep neural networks reveals drug resistance mechanisms in cancer.

Drug resistance remains one of the primary challenges in effective cancer therapy. In this study, we employed a deep neural network (DNN)-based transfer learning (TL) approach to predict drug response and uncover drug resistance mechanisms. We integrated gene expression, somatic mutation, and copy number aberration (CNA) data with drug response profiles using multi-omics integration (MI). We used the Genomics of Drug Sensitivity in Cancer (GDSC) data for training and incorporated drugs with same pathways into the training models. We then evaluated drug response predictions on independent in-vivo PDX Encyclopedia (PDX) and ex-vivo the Cancer Genome Atlas (TCGA) datasets. In addition, we conducted pathway enrichment analyses to elucidate the mechanisms underlying drug resistance for paclitaxel, 5-fluorouracil (5-FU), gemcitabine, and cetuximab. We also applied Fisher's exact test (FET) to assess potential associations between drug resistance and the presence of mutations or CNAs. Our pan-drug models outperformed other methods based on the area under the precision-recall curve (AUCPR). Our pathway enrichment analyses revealed LDHB-mediated pyruvate metabolism and FYN-mediated focal adhesion might have pivotal roles in paclitaxel resistance, while PINK1-mediated mitophagy might be critical in 5-FU resistance. In addition to transcriptional activation, FET suggested that CNAs in LDHB and PINK1 may also be associated with resistance to paclitaxel and 5-FU, respectively. Furthermore, enrichment results for paclitaxel and cetuximab indicated shared resistance mechanisms between the two drugs. Importantly, our findings are consistent with prior experimental studies, providing literature-based validation of our results. Overall, our DNN-based TL approach achieved strong predictive performance across PDX & TCGA datasets and enrichment analyses provided valuable biological insights into drug resistance mechanisms.

Humans↗

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

BIWT: a bioinformatics walkthrough for embedding spatial multiomics in agent-based models for virtual cells.

SUMMARY: Whereas transcriptomic and spatial profiling offer static snapshots of tissue structure, mechanistic models use biological rules to predict how tissues evolve. We present the BioInformatics WalkThrough (BIWT) software to directly initialize spatial agent-based models from single-cell and spatial molecular data. We demonstrate how initialization strategies affect tumor-immune dynamics and spatial clustering, positioning BIWT as a software suite to generate data-driven virtual cells representing both experimental and clinical contexts. AVAILABILITY AND IMPLEMENTATION: The BIWT software is available at https://github.com/PhysiCell-Tools/PhysiCell-Studio. The sample dataset for running the BIWT is available at https://zenodo.org/records/16365625. The code and instructions for reproducing the use case example is available at https://github.com/drbergman/BIWT-Paper.

Software↗