Single-Cell Virtual Cytometer allows user-friendly and versatile analysis and visualization of multimodal single cell RNAseq datasets

With the development of single-cell sequencing technology, it is a hot research topic to identify the cell types using single-cell sequencing data, and many single-cell clustering algorithms have been developed to study this issue. These methods capture partial information of single-cell sequencing data, and obtain the different performance on the same data set. Combining these different results into one can improve the accuracy and validity. Here, we proposed ECBN, Ensemble Clustering based on Bayesian Network. ECBN can ensemble several different results of state-of-the-art single cell clustering methods, such as Seurat, CIDR, SC3 and t-SNE+k-means, and generate a more optimal clustering result through Bayesian network. Experiments are carried on the 5 single cell data sets and compared with 4 individual single cell clustering methods and 3 integrative methods. The size of experiment data sets ranges from 822 to 3605 and the results show that our method can achieve good performance. Moreover, ECBN can also use the graphical regularization to lighten the limitation which is generated by the different basis results.

BackgroundThere exists a critical transition or tipping point during the complex biological process. Such critical transition is usually accompanied by the catastrophic consequences. Therefore, hunting for the tipping point or critical state is of significant importance to prevent or delay the occurrence of catastrophic consequences. However, predicting critical state based on the high-dimensional small sample data is a difficult problem, especially for single-cell expression data.ResultsIn this study, we propose the comprehensive neighbourhood-based perturbed mutual information (CPMI) method to detect the critical states of complex biological processes. The CPMI method takes into account the relationship between genes and neighbours, so as to reduce the noise and enhance the robustness. This method is applied to a simulated dataset and six real datasets, including an influenza dataset, two single-cell expression datasets and three bulk datasets. The method can not only successfully detect the tipping points, but also identify their dynamic network biomarkers (DNBs). In addition, the discovery of transcription factors (TFs) which can regulate DNB genes and nondifferential ‘dark genes’ validates the effectiveness of our method. The numerical simulation verifies that the CPMI method is robust under different noise strengths and is superior to the existing methods on identifying the critical states.ConclusionsIn conclusion, we propose a robust computational method, i.e., CPMI, which is applicable in both the bulk and single cell datasets. The CPMI method holds great potential in providing the early warning signals for complex biological processes and enabling early disease diagnosis.

Large single-cell RNA datasets have contributed to unprecedented biological insight. Often, these take the form of cell atlases and serve as a reference for automating cell labeling of newly sequenced samples. Yet, classification algorithms have lacked the capacity to accurately annotate cells, particularly in complex datasets. Here we present SIMS (Scalable, Interpretable Machine Learning for Single-Cell), an end-to-end data-efficient machine learning pipeline for discrete classification of single-cell data that can be applied to new datasets with minimal coding. We benchmarked SIMS against common single-cell label transfer tools and demonstrated that it performs as well or better than state of the art algorithms. We then use SIMS to classify cells in one of the most complex tissues: the brain. We show that SIMS classifies cells of the adult cerebral cortex and hippocampus at a remarkably high accuracy. This accuracy is maintained in trans-sample label transfers of the adult human cerebral cortex. We then apply SIMS to classify cells in the developing brain and demonstrate a high level of accuracy at predicting neuronal subtypes, even in periods of fate refinement, shedding light on genetic changes affecting specific cell types across development. Finally, we apply SIMS to single cell datasets of cortical organoids to predict cell identities and unveil genetic variations between cell lines. SIMS identifies cell-line differences and misannotated cell lineages in human cortical organoids derived from different pluripotent stem cell lines. When cell types are obscured by stress signals, label transfer from primary tissue improves the accuracy of cortical organoid annotations, serving as a reliable ground truth. Altogether, we show that SIMS is a versatile and robust tool for cell-type classification from single-cell datasets.

Single-cell sequencing technologies that simultaneously generate multimodal cellular profiles present opportunities for improved understanding of cell heterogeneity in tissues. How the multimodal information can be integrated to obtain a common cell type identification, however, poses a computational challenge. Multilayer graphs provide a natural representation of multi-omic single-cell sequencing datasets, and finding cell clusters may be understood as a multilayer graph partition problem. We introduce two spectral algorithms on multilayer graphs, spectral clustering on multilayer graphs and the weighted locally linear (WLL) method, to cluster cells in multi-omic single-cell sequencing datasets. We connect these algorithms through a unifying mathematical framework that represents each layer using a Hamiltonian operator and a mixture of its eigenstates to integrate the multiple graph layers, demonstrating in the process that the WLL method is a rigorous multilayer spectral graph theoretic reformulation of the popular Seurat weighted nearest neighbor (WNN) algorithm. Implementing our algorithms and applying them to a CITE-seq dataset of cord blood mononuclear cells yields results similar to the Seurat WNN analysis. Our work thus extends spectral methods to multimodal single-cell data analysis. The code used in this study can be found at https://github.com/jssong-lab/sc-spectrum. All public data used in the article are accurately cited and described in Materials and Methods and in Supplementary Information. Supplementary data are available at Bioinformatics online.

High-throughput single-cell sequencing technologies hold tremendous potential for defining cell types in an unbiased fashion using gene expression and epigenomic state. A key challenge in realizing this potential is integrating single-cell datasets from multiple protocols, biological contexts, and data modalities into a joint definition of cellular identity. We previously developed an approach, called linked inference of genomic experimental relationships (LIGER), that uses integrative nonnegative matrix factorization to address this challenge. Here, we provide a step-by-step protocol for using LIGER to jointly define cell types from multiple single-cell datasets. The main stages of the protocol are data preprocessing and normalization, joint factorization, quantile normalization and joint clustering, and visualization. We describe how to jointly define cell types from single-cell RNA-seq (scRNA-seq) and single-nucleus ATAC-seq (snATAC-seq) data, but similar steps apply across a wide range of other settings and data types, including cross-species analysis, single-nucleus DNA methylation, and spatial transcriptomics. Our protocol contains examples of expected results, describes common pitfalls, and relies only on our freely available, open-source R implementation of LIGER. We also provide R Markdown tutorials showing the outputs from each individual code segment. The analysis process can be performed in 1-4 h, depending on dataset size, and assumes no specialized bioinformatics training.

Multiple myeloma (MM) is a hematological malignancy of aberrant clonal plasma cells that reside within the bone marrow (BM).1 The disease course differs from other BM malignancies essentially in 2 features. First, there are 2 precursor stages, monoclonal gammopathy of undetermined significance (MGUS) and smoldering MM (SMM), that can transition into MM over time.2 Second, once clinical MM develops, the disease remains largely incurable, and despite significant therapeutic improvements, relapse and refractoriness usually cannot be prevented. The interactions between MM cells and the BM microenvironment (BM-ME) are an area of particularly intense research, as tumor evasion and suppression of the host immune system constellate main factors of MM progression.3 Single-cell sequencing technologies that have emerged over the last years have the potential to significantly advance the field because they enable the evaluation of alterations in cell numbers and states as well as interactions between MM cells and the BM-ME. Recent studies have applied single-cell techniques at different precursor and MM stages to determine the comprehensive changes in the BM-ME and to identify mechanisms that foster oncogenesis.4–11 This article briefly summarizes these studies and proposes how the dissection of the BM-ME on a single-cell level can improve our understanding of MM pathogenesis, thereby advancing prognostication and the therapeutic landscape. We would like to emphasize that single-cell analyses are a rapidly expanding field and the number of published studies investigating the BM-ME in MM and its precursors is likely to grow exponentially in the next few years given the increased interest in and access to single-cell technologies. Thus, this perspective does not cover all available articles. THE IMMUNE MICROENVIRONMENT IN THE EARLY EVOLUTION OF MYELOMA Progression from precursor stages to clinical MM is highly heterogeneous, with some patients progressing quickly, while the others remain stable for decades.2 Treatment is hence not justified solely on the premise of a precursor diagnosis. Hence, recent research has focused on identifying SMM patients, who would benefit from early therapeutic interventions. Current risk classifiers are mainly based on clinical parameters and tumor aberrations (translocation t[4;14], t[14;16], gain1q and/or del13q), yet are only able to capture up to ~60% of patients who progress to MM within 2 years.12 Further molecular characterization of the BM-ME could improve the accuracy of prognostication and also pave the way for better therapeutic avenues. In that regard, Zavidij et al8 performed single-cell RNA sequencing (scRNA-seq) of the immune-ME from 5 MGUS, 11 SMM and 7 MM patients compared with 9 healthy individuals. The authors show that substantial alterations of the immune-ME are already present at the MGUS stage. These include increased populations of natural killer (NK) cells, T cells, particularly Tregs, and nonclassical CD16+ monocytes. CD14+ monocytes showed dysregulated expression of major histocompatibility complex (MHC) type II genes, which resulted in T-cell suppression in in vitro cultures. The progression from MGUS to SMM was associated with the loss of granzyme K+ memory cytotoxic T cells leading to reduced immunosurveillance in in vivo models. Symptomatic MM was characterized by an increased INF-alpha response across all immune cell types, which has been shown to promote immunosuppression, favoring expansion of MM cells. Similarly, using scRNA-seq and mass cytometry, Bailur et al11 showed that precursor stages already harbor BM-ME alterations compared with healthy donors, including early changes in NK and myeloid cells as well as increased terminal effector differentiation and enrichment of stem-like T cells in MGUS. Another scRNA-seq study with 8 MGUS, 7 SMM, and 10 MM patients showed substantial alterations in the immune-ME of MM compared with its precursor stages.5 Symptomatic MM was enriched for CD14+ and CD16+ monocyte populations, memory B and CD8 effector cells compared with early stages with a transitional decrease of CD4-positive T cells and MAIT cells. However, there was no differential expression of exhaustion markers such as PD1, TIGIT, and LAG-3 when comparing stages. Importantly, cell proportions were heterogeneous even across the same disease stage, emphasizing the need for larger datasets and also underscoring the existence of an individual immune phenotype, which is likely influenced by other factors than the aberrant plasma cells. THE IMMUNE MICROENVIRONMENT DURING MYELOMA THERAPY AND IN RELAPSING DISEASE Established therapeutic strategies yield high complete remission rates, but disease relapse will ensue in the vast majority of patients. Yet, the disease course remains very heterogeneous with some patients relapsing within months of treatment initiation and others experiencing deep remission for >10 years. This observation has led to the hypothesis that MM cells can enter a dormant state with regrowth over time when conditions become permissive, particularly when immune surveillance falters.13 Although many MM therapies try to stimulate the immune-ME to fight the disease, it has been also shown that therapy causes immunosuppression with ensuing cytopenias and hypogammaglobulinemia, emphasizing that the balance of a stimulated or dysfunctioning immune-ME is very delicate. Deciphering the alterations within the immune-ME during therapy will hence be challenging. Tirier et al6 investigated 20 relapsed/refractory MM patients prior and post salvage therapy and compared them to 8 healthy donors. Applying scRNA-seq to the immune-ME and plasma cells, the authors analyzed both compositional changes and alterations in cell–cell interactions. In this population, which was heavily enriched for gain1q, the investigators found depletion of CD4+/CD8+ naïve and CD4+ memory cells with enrichment of CD14+/CD16+ monocytes, effector T-cell populations (CD8+ memory and cytotoxic cells) as well as gamma/delta T cells with an increase of inflammatory markers in the BM that were primarily secreted by CD14+ monocytes and MM cells. In a similar attempt to understand co-evolution of MM cells and their immune-ME, Liu et al4 described longitudinal alterations during disease progression in 14 MM patients. They report that genetic alterations tend to be associated with distinct immune cells clusters, for example, patients with translocation t(11;14) showed separate T-cell clusters with upregulation of lysine methyltransferase genes KMT2A and KMT2C in CD8+ T cells. Other studies addressed the role of the immune-ME in early relapse patients. Pilcher et al7 performed scRNA-seq on the immune-ME in patients with rapidly progressing disease (<18 months) compared with those without progression within 4 years of follow-up. Early relapse patients had significantly higher numbers of exhausted CD8+ T cells (GZMK+ and TIGIT) with decreased expression of cytotoxic markers (PRF1, GZMB, and GNLY). Similar to the study by Tirier et al, there were also alterations within the monocyte/macrophage compartment with an increase of M2 macrophages in rapid progressors. Yao et al combined scRNA-seq with an additional capture of surface markers (Cellular Indexing of Transcriptomes and Epitopes by Sequencing, CITE-Seq) and the mass cytometry approach cytometry by time of flight (CyTOF) to elucidate the biology of early relapse.9 Stratifying 18 MM patients by international staging system (ISS) and by time to progression (<6 months for rapid progressors or 6 months to 5 years for slow progressors), they found aggressive disease (ISS 3) to be associated with a decrease in CD4+ T cells and the CD4/CD8 ratio. Last but not least, recent studies investigated alterations of the BM-ME during immunotherapies, as those heavily rely on a functional immune system for optimal efficacy. Adams III et al determined the effect of the CD38-antibody Daratumumab on immune cells in the BM and peripheral blood (PB) in relapsed MM patients.10 Daratumumab depleted NK cells and reduced the amount of CD38+ basophils. As seen in the previous studies, immune profiling differed between responders and nonresponders with higher granzyme B expression in CD8+ T cells of responders, suggesting treatment induced T-cell activation with increased killing capacity. Friedrich et al determined the immunological single-cell mechanisms of resistance to bispecific BCMAxCD3 T-cell engager (TCE) therapy, a novel immunotherapy leading to unprecedented response rates in refractory MM.14 They show that TCEs lead to clonal expansion of immune cell subsets, particularly effector CD8+ T cells and that primary or acquired failure to TCE therapy is associated with exhaustion of CD8+ T cells. Importantly, treatment failure or loss of response was not solely associated with immune-ME alterations but also MM intrinsic adaptations, such as loss of the target epitope and MHC class I protein. Taken together, current research is rapidly progressing to identifying alterations within the immune-ME of particularly aggressive disease, as these patients are in most need of novel therapeutic approaches. Single-cell sequencing studies so far appear to uniformly show dysregulation of T cells, particularly cytotoxic/effector T cells with increase of exhaustion markers in progressing patients. Furthermore, there appear to be alterations leading to a dysfunctional state in the monocytes/macrophages compartment in more aggressive MM. CHALLENGES AND FUTURE PERSPECTIVES Single-cell techniques have the potential to comprehensively dissect disease mechanisms within the immune-ME and their relationship with MM cells, thereby offering a path to truly individualized medicine leading to improved prognostic measure and therapeutic avenues. Challenges remain in unifying cell subset annotations to better compare studies. The encountered heterogeneity between patients could be accounted for by increasing sample sizes and by stratifying patients based on the parameters such as age, genetic profiles of MM, and treatment received. Age has shown to have a significant impact on immune function and is manifested at multiple levels including reduced production of B and T cells and diminished function of mature lymphocytes.15 It is hence not surprising that increased age has repeatedly shown to be an adverse risk factor for clinical outcome in MM, which is likely not only due to the higher prevalence of comorbidities in this population.16 The notion that molecular subgroups of MM can differ in the surrounding BM-ME has been reported long before the onset of single-cell technologies. For example, patients of the low bone disease molecular subgroup barely show osteolytic lesions, suggesting that this subset spares the stimulation of osteoclasts and the resulting bone resorption.17 The recently discovered associations of an immune-ME signature and +1q MM and the distinct clustering of immune cells in patients with translocation t(11;14) further corroborate the hypothesis that the MM genotype and phenotype influences its surrounding or vice versa. Hence, analyzing MM in its BM-ME on a single-cell level will offer the unique opportunity to associate alterations in the ME to intrinsic MM cell features, which could revolutionize our understanding of MM biology. Treatment has a significant effect on the immune-ME, and can lead to alterations in cell populations and expression patterns, which will need to be taken into account when analyzing single-cell data.10,18,19 These alterations are associated with immune cell cytotoxicity, exhaustion, and senescence, which can influence the further disease course.20 Conversely, comprehensive immune profiling before therapy could determine which treatment modalities would yield best responses. The use of single-cell techniques before and during therapy would enable us to dissect mechanisms of response versus resistance and could tailor therapeutic approaches based on the patients' individualized immune profile. Additional challenges to single-cell technologies are based on the current limitations of these techniques and are unlikely to be overcome by merely increasing the number of samples. This includes the limited number of analyzed cells due to the high costs. As an alternative or complementary approach, multidimensional flow cytometry allows for a cost-effective assessment of immune-phenotypes in the BM-ME and yields prognostic information in MM and its precursor stages.19,21,22 A particular hindrance to single-cell technologies is the lack of spatial resolution as they are performed on BM aspirates or PB. Although the current algorithms can predict interactions between cell populations based on the surface markers, the BM-ME architecture remains elusive. Furthermore, aspirates contain very small numbers of adherent cells, such as stromal cells, osteoblasts, and osteoclasts. Hence, to determine the structure and composition of the BM and its relationship with MM in great detail, it will be important to combine single-cell technologies with multiplex immunohistochemistry studies.23 Of note, BM-ME single-cell data are usually derived from a random BM site and might not be representative of the whole disease process. This is particularly important for MM, as the vast majority of patients present with focal lesions where genetic profiles of MM cells have been shown to be distinct from the random BM site.24,25 First reports suggest that this also holds true for the BM-ME at these sites.26,27 There is growing awareness that MM cell heterogeneity is, to at least some degree, reflected by circulating tumor cells, which are thought to originate from all affected MM sites and have prognostic impact.22 Thus, paired profiling of the BM and PB could unravel alterations that are not evident at the random BM site. A summary of the anticipated comprehensive analysis of MM and its BM-ME in the future is presented in Figure 1.Figure 1.: Summary of the anticipated future comprehensive analysis of myeloma and its microenvironment at a single-cell level.In conclusion, single-cell technologies have already influenced the MM field by identifying mechanisms underlying treatment resistance and they have the potential to further advance our current understanding of risk classification, prognostication, and individualized treatment approaches. While high costs currently limit their application, it is anticipated that technical advances and broader use will significantly reduce the costs and allow for larger sample collections. Hence, comprehensive and longitudinal profiling of the immune-ME using multiomic single-cell approaches with integration of the spatial BM architecture and genomic profiling of MM cells could soon be a reality for our MM patients and will eventually lead to improved clinical outcomes and cure. ACKNOWLEDGMENTS As a result of the space constraints, we apologize that we were unable to cite all our colleagues that have made an impact in defining our current knowledge on this topic. DISCLOSURES NW was supported by the Dietmar-Hopp Foundation. The authors declare no conflicts of interest. SOURCES OF FUNDING NW was supported by the Dietmar-Hopp Foundation.

In general, the ‘small sample (n)-large feature ( Open image in new window )’ problem of bioinformatics, image analysis, high throughput molecular screening, astronomy, and other high dimensional applications makes the features highly collinear. In this context, the paper presents a new feature extraction algorithm to address this ‘large Open image in new window small n’ issue associated with multimodal data sets. The proposed algorithm judiciously integrates the concept of both regularization and shrinkage with canonical correlation analysis to extract important features. To deal with the singularity problem, the proposed method increases the diagonal elements of covariance matrices by using regularization parameters, while the off-diagonal elements are decreased by shrinkage coefficients. The concept of hypercuboid equivalence partition matrix of rough hypercuboid approach is used to compute both significance and relevance measures of a feature. The importance of the proposed algorithm over other existing methods is established extensively on real life multimodal omics data set.

BackgroundAdvancements in genomic sequencing continually improve personalized medicine, and recent breakthroughs generate multimodal data on a cellular level. We introduce MOSCATO, a technique for selecting features across multimodal single-cell datasets that relate to clinical outcomes. We summarize the single-cell data using tensors and perform regularized tensor regression to return clinically-associated variable sets for each ‘omic’ type.ResultsRobustness was assessed over simulations based on available single-cell simulation methods, and applicability was assessed through an example using CITE-seq data to detect genes associated with leukemia. We find that MOSCATO performs favorably in selecting network features while also shown to be applicable to real multimodal single-cell data.ConclusionsMOSCATO is a useful analytical technique for supervised feature selection in multimodal single-cell data. The flexibility of our approach enables future extensions on distributional assumptions and covariate adjustments.

Accumulating evidence has suggested that intervertebral disc degeneration (IVDD) serves as a health problem deserving attention from worldwide, owing to the low back pain and even disability it causes. Cellular senescence, an irreversible process of terminal cell cycle arrest, plays an essential role in the degeneration of nucleus pulposus (NP) cells and IVDD with an unknown underlying mechanism. Here, we have explored the senescence signature in IVDD through high dimensional weighted gene co-expression network analysis (hdWGCNA) algorithm in two single cell datasets (PRJCA014236 and GSE244889) to construct a novel signature called NP_Senescence, which accurately identifies the patients with IVDD. The efficacy and accuracy were validated in various external datasets, including two expression array datasets (GSE70362 and GSE34095) and one single cell dataset (GSE230809). Additionally, the accuracy of the specific model was also verified with the senescent cell identification (SenCID) algorithm. IVDD patients with a high NP_Senescence score exhibited the characteristics of compensated activation of hyaluronic acid monomers synthesis and abnormal status of DNA damage repair through single-cell Flux Estimation Analysis (scFEA) algorithm. Furthermore, the NP_Senescence was negatively associated with the level of AMP and Glucose-1-phosphate, while positively associated with the level of β-alanine and UDP-glucuronic acid. In conclusion, we constructed a novel disc-specific scoring model NP_Senescence by comprehensively analyzing the senescence-related genes of IVDD, which can be used for patient early diagnosis and treatment, as well as the identification of metabolic characteristics.

Atypical Teratoid Rhabdoid Tumors (ATRTs) are pediatric malignancies that arise primarily in young children under the age of 3. While ATRTs are known to segregate into three epigenetically defined subgroups, termed SHH, TYR, and MYC, little is known about the cellular contexts of how these deadly tumors arise and gain tumorigenic properties. We hypothesized that this early onset of disease reflects tumorigenic events in utero. Indeed, previous research has demonstrated maintenance of embryogenesis programs in other pediatric brain tumors. To characterize the cellular contexts that give rise to ATRTs, we curated 12 single cell and single nuclei RNAseq datasets of human and mouse embryogenesis and developed an atlas and well as a pipeline to match tumors to their most similar embryogenic landmark and timepoint, based on the published cell type identification tool SingleR. We applied this pipeline to two ATRTs single cell datasets generated on the 10X genomics platform: one from primary human samples (N = 33), and one from a tamoxifen inducible ATRT mouse model (N = 8). To validate the accuracy of our pipeline, we first examined the signatures from our ATRT mouse model. Our pipeline accurately predicted cell types and timepoint within the induction window of E6.5-8.5, suggesting that indeed tumors transcriptionally resemble their cellular contexts. Results on both human and mouse ATRTs reveals that ATRTs resemble two distinct developmental trajectories: one resembling a gradient of neuronal differentiation from neuroepithelial (NES, SOX2, OTX2) to radial glial cell (TTYH1, ASCL1, FAPB7) types, and a second trajectory characterized by distinctly mesenchymal signatures (COL1A2, PDGFRB). Single cell GSEA showed enrichment of pathways involved in radial glial and mesenchymal signalling, such as WNT, SHH, and PDGFRB signalling. Findings were validated through marker gene immunohistochemistry and spatial transcriptomics. Taken together, our results demonstrate that ATRTs transcriptionally resemble the developmental contexts of which they arise from. By identifying cellular origins of tumorigenesis, we can exploit this novel information to discover potential therapeutic avenues. Citation Format: Fupan Yao, Dean Popovski, Andrew Bondoc, James Rutka, Trevor Pugh, Annie Huang. Atypical teratoid rhabdoid tumor development mirrors developmental programs [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6445.

IntroductionChronic immunopathology contributes to the development of heart failure after a myocardial infarction. Both T and B cells of the adaptive immune system are present in the myocardium and have been suggested to be involved in post-MI immunopathology. MethodsWe analyzed the B and T cell populations isolated from previously published single cell RNA-sequencing data sets (PMID: 32130914, PMID: 35948637, PMID: 32971526 and PMID: 35926050), of the mouse and human heart, using differential expression analysis, functional enrichment analysis, gene regulatory inferences, and integration with autoimmune and cardiovascular GWAS. ResultsAlready at baseline, mature effector B and T cells are present in the human and mouse heart, having increased activity in transcription factors maintaining tolerance (e.g. DEAF1, JDP2, SPI-B). Following MI, T cells upregulate pro-inflammatory transcript levels (e.g. Cd11, Gzmk, Prf1), while B cells upregulate activation markers (e.g. Il6, Il1rn, Ccl6) and collagen (e.g. Col5a2, Col4a1, Col1a2). Importantly, pro-inflammatory and fibrotic transcription factors (e.g. NFKB1, CREM, REL) remain active in T cells, while B cells maintain elevated activity in transcription factors related to immunoglobulin production (e.g. ERG, REL) in both mouse and human post-MI hearts. Notably, genes differentially expressed in post-MI T and B cells are associated with cardiovascular and autoimmune disease. ConclusionThese findings highlight the varied and time-dependent dynamic roles of post-MI T and B cells. They appear ready-to-go and are activated immediately after MI, thus participate in the acute wound healing response. However, they subsequently remain in a state of pro-inflammatory activation contributing to persistent immunopathology.

Recent developments of video and sensing technology have led to large volumes of digital media data. Current media production relies on videos from the principal camera together with a wide variety of heterogeneous source of supporting data [photos, light detection and ranging point clouds, witness video camera, high dynamic range imaging, and depth imagery]. Registration of visual data acquired from various 2D and 3D sensing modalities is challenging because current matching and registration methods are not appropriate due to differences in structure, format, and noise characteristics for multimodal data. A combined 2D/3D visualization of this registered data allows an integrated overview of the entire data set. For such a visualization, a Web-based context presents several advantages. In this paper, we propose a unified framework for registration and visualization of this type of visual media data. A new feature description and matching method is proposed, adaptively considering local geometry, semiglobal geometry, and color information in the scene for more robust registration. The resulting registered 2D/3D multimodal visual data are too large to be downloaded and viewed directly via the Web browser, while maintaining an acceptable user experience. Thus, we employ hierarchical techniques for compression and restructuring to enable efficient transmission and visualization over the Web, leading to interactive visualization as registered point clouds, 2D images, and videos in the browser, improving on the current state-of-the-art techniques for Web-based visualization of big media data. This is the first unified 3D Web-based visualization of multimodal visual media production data sets. The proposed pipeline is tested on big multimodal data set typical of film and broadcast production, which are made publicly available. The&lt;br/&gt;proposed feature description method shows two times higher precision of feature matching and more stable registration performance than existing 3D feature descriptors.

Many neurodegenerative diseases (such as Alzheimer’s disease) and neuropsychiatric disorders (such as schizophrenia) are still poorly understood. While incorporating multimodal datasets (structural and functional) would provide a more complete picture of these neurological conditions, existing studies already struggle with two key challenges even when a single data modality is used: data scarcity and high dimensionality. These problems make machine learning models overfit easily on neuroimaging datasets and result in the generation of spurious biomarkers. Coupled with issues such as inter-scanner variability and disease heterogeneity, biomarker discovery remains an onerous task as reproducible biomarkers for these neurological conditions have proven to be elusive. Overcoming these obstacles would give us a better understanding of these diseases and possibly lead to the discovery of new cures. One emerging trend in the area of biomarker discovery is the use of deep learning models. Despite its perceived complexity, one relatively unexplored aspect of deep learning architectures is the flexibility they afford for modelling multimodal datasets, in terms of architectural design and construction of the loss function. In this thesis, we propose several deep learning architectures customised for the idiosyncrasies of brain imaging datasets and demonstrate how they can produce reliable biomarkers. Firstly, the challenges of high dimensionality and data scarcity in brain imaging datasets are investigated. Existing approaches alleviated the issue by removing less important nodes from the neural network architecture, including the input features. However, it is common to have sites with very small datasets (< 100) where such pruning-based approaches are no longer effective. Thus, we propose an alternative approach based on data harmonisation and semi-supervised learning that tackles this issue by incorporating more data (on the same disease, but collected from other sites) into the analysis while addressing issues such as site differences and labelling inconsistencies. We demonstrated that existing works on biomarker discovery produce biomarkers that could be biased towards the largest site, especially when data imbalance exists across sites. On the other hand, our proposed architecture provides a technique to arrive at site-invariant biomarkers, making it also possible to reveal site-specific biomarkers more clearly. Secondly, a persistent issue with existing machine learning models is their poor generalisability on neuroimaging datasets. This is often hypothesised to be caused by disease heterogeneity. Furthermore, while some sites are carefully curated such that they represent a homogeneous subgroup, many sites still contain a heterogeneous mix of patients, warranting a need for techniques that can capture such variability. Our proposed approach uses both the brain graph and population graph approach to model structural and functional brain networks. It introduces an additional fine-tuning step that adds clustering information into the construction of the population graph, based on the clusters obtained from the brain graph embeddings. This makes it possible to produce subtype-specific biomarkers. Thirdly, we explore the possibility of combining both multimodal neuroimaging data and multi-omics datasets. Structural and functional neuroimaging data provides whole-brain, macroscale information but the limited spatial resolution leaves out microscale information. These are better captured via multi-omics data, but such multimodal and multi-scale imaging genetics research has been relatively under-explored as it further exacerbates the above-mentioned problem of model overfitting. To address this, we propose a scalable deep neural network architecture based on the attention mechanism that allows a myriad of omics modalities to be incorporated into the analysis. We demonstrate how combining imaging and genetics data leads to better model performance and go further to show how relative importance of the modalities involved in the analysis can be determined. Overall, these works demonstrated how existing biomarker discovery approaches are limited to a generic, class-wide view of the disease that could be biased especially for imbalanced multi-site datasets. These problems are addressed by careful design of deep learning architectures (graph neural networks with data harmonisation, semi-supervised learning, clustering and the attention mechanism) to take care of issues unique to biomedical datasets, leading to better model performance and producing richer biomarkers in the form of site-specific, site-invariant, subtype-specific and imaging genetics biomarkers. These biomarkers pave the way towards more effective treatments for these complex neurological conditions.

Automatic depression classification from multimodal input data is a challenging task. Modern methods use paralinguistic information such as audio and video signals. Using linguistic information such as speech signals and text data for depression classification is a complicated task in deep learning models. Best audio and video features are built to produce a dependable depression classification system. Textual signals related to depression classification are analyzed using text-based content data. Moreover, to increase the achievements of the depression classification system, audio, visual, and text descriptors are used. So, a deep learning-based depression classification model is developed to detect the person with depression from multimodal data. The EEG signal, Speech signal, video, and text are gathered from standard databases. Four stages of feature extraction take place. In the first stage, the features from the decomposed EEG signals are attained by the empirical mode decomposition (EMD) method, and features are extracted by means of linear and nonlinear feature extraction. In the second stage, the spectral features of the speech signals from the Mel-frequency cepstral coefficients (MFCC) are extracted. In the third stage, the facial texture features from the input video are extracted. In the fourth stage of feature extraction, the input text data are pre-processed, and from the pre-processed data, the textual features are extracted by using the Transformer Net. All four sets of features are optimally selected and combined with the optimal weights to get the weighted fused features using the enhanced mountaineering team-based optimization algorithm (EMTOA). The optimal weighted fused features are finally given to the hybrid attention-based dilated network (HADN). The HDAN is developed by combining temporal convolutional network (TCN) with bidirectional long short-term memory (Bi-LSTM). The parameters in the HDAN are optimized with the assistance of the developed EMTOA algorithm. At last, the classified output of depression is obtained from the HDAN. The efficiency of the developed deep learning HDAN is validated by comparing it with various traditional classification models.

Revealing the underlying cell-state landscape from single-cell data requires overcoming the critical obstacles of batch integration, denoising, and dimensionality reduction. Here, we present CONCORD, a unified framework that simultaneously addresses these challenges within a single self-supervised model. At its core, CONCORD implements a unified probabilistic sampling strategy that corrects batch effects via dataset-aware sampling and enhances biological resolution through hard-negative sampling. Remarkably, using only a minimalist neural network with a single hidden layer and contrastive learning, CONCORD surpasses state-of-the-art performance without relying on deep architectures, auxiliary losses, or external supervision. It seamlessly integrates data across batches, technologies, and even species to generate high-resolution cell atlases. The resulting latent representations are denoised and biologically meaningful—capturing gene co-expression programs, revealing detailed lineage trajectories, and preserving both local geometric relationships and global topological structures. We demonstrate CONCORD’s broad applicability across diverse datasets, establishing it as a general-purpose framework for learning unified, high-fidelity representations of cellular identity and dynamics.