Nonsmooth nonnegative matrix factorization (nsNMF)

Conserved methylation signatures associate with the tumor immune microenvironment and immunotherapy response

BACKGROUND

Aberrant DNA methylation is a major characteristic of cancer genomes. It remains unclear which biological processes determine epigenetic reprogramming and how these processes influence the variants in the cancer methylome, which can further impact cancer phenotypes.

METHODS

We performed pairwise permutations of 381,900 loci in 569 paired DNA methylation profiles of cancer tissue and matched normal tissue from The Cancer Genome Atlas (TCGA) and defined conserved differentially methylated positions (DMPs) based on the resulting null distribution. Then, we derived independent methylation signatures from 2,465 cancer-only methylation profiles from the TCGA and 241 cell line-based methylation profiles from the Genomics of Drug Sensitivity in Cancer (GDSC) cohort using nonnegative matrix factorization (NMF). We correlated DNA methylation signatures with various clinical and biological features, including age, survival, cancer stage, tumor immune microenvironment factors, and immunotherapy response. We inferred the determinant genes of these methylation signatures by integrating genomic and transcriptomic data and evaluated the impact of these signatures on cancer phenotypes in independent bulk and single-cell RNA/methylome cohorts.

RESULTS

We identified 7,364 differentially methylated positions (2,969 Hyper-DMPs and 4,395 Hypo-DMPs) in nine cancer types from the TCGA. We subsequently retrieved three highly conserved, independent methylation signatures (Hyper-MS1, Hypo-MS1, and Hypo-MS4) from cancer tissues and cell lines based on these Hyper and Hypo-DMPs. Our data suggested that Hypo-MS4 activity predicts poor survival and is associated with immunotherapy response and distant tumor metastasis, and Hypo-MS4 activity is related to TP53 mutation and FOXA1 binding specificity. In addition, we demonstrated a correlation between the activities of Hypo-MS4 in cancer cells and the fractions of regulatory CD4 + T cells with the expression levels of immunological genes in the tumor immune microenvironment.

CONCLUSIONS

Our findings demonstrated that the methylation signatures of distinct biological processes are associated with immune activity in the cancer microenvironment and predict immunotherapy response.

Genome-scale CRISPR-Cas9 screening stratifies pancreatic cancer with distinct outcomes and immunotherapeutic efficacy

Pancreatic cancer (PC) was featured by dramatic heterogeneity and dismal outcomes. An ideal classification strategy capable of achieving risk stratification and individualized treatment is urgently needed to significantly improve prognosis. In this study, using the 105 prognostic cancer essential genes identified by genome-scale CRISPR-Cas9 screening and univariate Cox analysis, we established and verified three heterogeneous subtypes via non-negative matrix factorization (NMF) and nearest template prediction (NTP) algorithms in the TCGA-PAAD cohort (176 samples) and four multi-center cohorts (233 samples), respectively. Among them, C1 with the worst prognosis was enriched in immune-related pathways, possessed superior infiltration abundance of immune cells and immune checkpoint molecules expression, and might be most sensitive to immunotherapy. C3, owing a moderate prognosis, might be featured by proliferative biological function, and despite its highest immunogenicity, the defects in antigen processing and presentation ability coupled with barren immune environment render it ineffective for immunotherapy, while it had potential sensitivity to paclitaxel and methotrexate. Besides, C2 harbored the best prognosis and was characterized by metabolism-related functions. These results could deepen our understanding of PC molecular heterogeneity and provide a trustworthy reference for prognostic stratification management and precision medicine in clinical practice.

Frequency-specific directed interactions between whole-brain regions during sentence processing using multimodal stimulus

Neural oscillations subserve a broad range of speech processing and language comprehension functions. Using an electroencephalogram (EEG), we investigated the frequency-specific directed interactions between whole-brain regions while the participants processed Chinese sentences using different modality stimuli (i.e., auditory, visual, and audio-visual). The results indicate that low-frequency responses correspond to the process of information flow aggregation in primary sensory cortices in different modalities. Information flow dominated by high-frequency responses exhibited characteristics of bottom-up flow from left posterior temporal to left frontal regions. The network pattern of top-down information flowing out of the left frontal lobe was presented by the joint dominance of low- and high-frequency rhythms. Overall, our results suggest that the brain may be modality-independent when processing higher-order language information.

Temperament & Character account for brain functional connectivity at rest: A diathesis-stress model of functional dysregulation in psychosis

The human brain's resting-state functional connectivity (rsFC) provides stable trait-like measures of differences in the perceptual, cognitive, emotional, and social functioning of individuals. The rsFC of the prefrontal cortex is hypothesized to mediate a person's rational self-government, as is also measured by personality, so we tested whether its connectivity networks account for vulnerability to psychosis and related personality configurations. Young adults were recruited as outpatients or controls from the same communities around psychiatric clinics. Healthy controls (n = 30) and clinically stable outpatients with bipolar disorder (n = 35) or schizophrenia (n = 27) were diagnosed by structured interviews, and then were assessed with standardized protocols of the Human Connectome Project. Data-driven clustering identified five groups of patients with distinct patterns of rsFC regardless of diagnosis. These groups were distinguished by rsFC networks that regulate specific biopsychosocial aspects of psychosis: sensory hypersensitivity, negative emotional balance, impaired attentional control, avolition, and social mistrust. The rsFc group differences were validated by independent measures of white matter microstructure, personality, and clinical features not used to identify the subjects. We confirmed that each connectivity group was organized by differential collaborative interactions among six prefrontal and eight other automatically-coactivated networks. The temperament and character traits of the members of these groups strongly accounted for the differences in rsFC between groups, indicating that configurations of rsFC are internal representations of personality organization. These representations involve weakly self-regulated emotional drives of fear, irrational desire, and mistrust, which predispose to psychopathology. However, stable outpatients with different diagnoses (bipolar or schizophrenic psychoses) were highly similar in rsFC and personality. This supports a diathesis-stress model in which different complex adaptive systems regulate predisposition (which is similar in stable outpatients despite diagnosis) and stress-induced clinical dysfunction (which differs by diagnosis).

Transcriptional space-time mapping identifies concerted immune and stromal cell patterns and gene programs in wound healing and cancer

Tissue repair responses in metazoans are highly coordinated by different cell types over space and time. However, comprehensive single-cell-based characterization covering this coordination is lacking. Here, we captured transcriptional states of single cells over space and time during skin wound closure, revealing choreographed gene-expression profiles. We identified shared space-time patterns of cellular and gene program enrichment, which we call multicellular "movements" spanning multiple cell types. We validated some of the discovered space-time movements using large-volume imaging of cleared wounds and demonstrated the value of this analysis to predict "sender" and "receiver" gene programs in macrophages and fibroblasts. Finally, we tested the hypothesis that tumors are like "wounds that never heal" and found conserved wound healing movements in mouse melanoma and colorectal tumor models, as well as human tumor samples, revealing fundamental multicellular units of tissue biology for integrative studies.

Integrated Proteomics and Single-Cell Mass Cytometry Analysis Dissects the Immune Landscape of Ankylosing Spondylitis

Mass cytometry is a powerful single-cell technology widely adopted to depict immune cell heterogeneity in different contexts. However, this method is only capable of examining several dozens of proteins simultaneously and requires a prior knowledge of the markers to be analyzed. Here we propose that the integration of mass cytometry with shot-gun proteomics may serve as a valuable tool to achieve an in-depth understanding of the immune system. By implementing such a strategy, we investigated the immune landscape of ankylosing spondylitis (AS), a chronic inflammatory arthritis with unclear etiology. The proteome alteration in peripheral blood mononuclear cells (PBMCs) was investigated by quantitative proteomics, and then mass cytometry analysis was conducted to decipher the immunome by considering the signaling molecules identified with differential expression by proteomics. As a result, we identified a wide spectrum of proteins dysregulated in AS, e.g., upregulation of glycolytic enzymes, downregulation of lipid transporters, and dysregulation of chemokine signaling molecules involved in proinflammatory cytokine production and leucocyte migration. Moreover, the single-cell analysis showed the upregulation of chemokine signaling regulators in subclusters of both innate and adaptive immune cells in AS. In addition, correlation analysis unveiled the interplay among Phenograph-identified subclusters of monocytes, CD4+ T cells, and CD8+ T cells. Taken together, our findings demonstrated that the integration of mass spectrometry-based proteomics and single-cell mass cytometry may serve as a useful tool to reveal clinically relevant information regarding useful targets and cellular phenotypes that could be further exploited to develop novel therapeutic strategies.

Molecular subtypes of ALS are associated with differences in patient prognosis

Amyotrophic Lateral Sclerosis (ALS) is a neurodegenerative disease with poorly understood clinical heterogeneity, underscored by significant differences in patient age at onset, symptom progression, therapeutic response, disease duration, and comorbidity presentation. We perform a patient stratification analysis to better understand the variability in ALS pathology, utilizing postmortem frontal and motor cortex transcriptomes derived from 208 patients. Building on the emerging role of transposable element (TE) expression in ALS, we consider locus-specific TEs as distinct molecular features during stratification. Here, we identify three unique molecular subtypes in this ALS cohort, with significant differences in patient survival. These results suggest independent disease mechanisms drive some of the clinical heterogeneity in ALS.

Disease patterns of coronary heart disease and type 2 diabetes harbored distinct and shared genetic architecture

BACKGROUND

Coronary heart disease (CHD) and type 2 diabetes (T2D) are two complex diseases with complex interrelationships. However, the genetic architecture of the two diseases is often studied independently by the individual single-nucleotide polymorphism (SNP) approach. Here, we presented a genotypic-phenotypic framework for deciphering the genetic architecture underlying the disease patterns of CHD and T2D.

METHOD

A data-driven SNP-set approach was performed in a genome-wide association study consisting of subpopulations with different disease patterns of CHD and T2D (comorbidity, CHD without T2D, T2D without CHD and all none). We applied nonsmooth nonnegative matrix factorization (nsNMF) clustering to generate SNP sets interacting the information of SNP and subject. Relationships between SNP sets and phenotype sets harboring different disease patterns were then assessed, and we further co-clustered the SNP sets into a genetic network to topologically elucidate the genetic architecture composed of SNP sets.

RESULTS

We identified 23 non-identical SNP sets with significant association with CHD or T2D (SNP-set based association test, P < 3.70 × [Formula: see text]). Among them, disease patterns involving CHD and T2D were related to distinct SNP sets (Hypergeometric test, P < 2.17 × [Formula: see text]). Accordingly, numerous genes (e.g., KLKs, GRM8, SHANK2) and pathways (e.g., fatty acid metabolism) were diversely implicated in different subtypes and related pathophysiological processes. Finally, we showed that the genetic architecture for disease patterns of CHD and T2D was composed of disjoint genetic networks (heterogeneity), with common genes contributing to it (pleiotropy).

CONCLUSION

The SNP-set approach deciphered the complexity of both genotype and phenotype as well as their complex relationships. Different disease patterns of CHD and T2D share distinct genetic architectures, for which lipid metabolism related to fibrosis may be an atherogenic pathway that is specifically activated by diabetes. Our findings provide new insights for exploring new biological pathways.

Multiple Graph Adaptive Regularized Semi-Supervised Nonnegative Matrix Factorization with Sparse Constraint for Data Representation

Multiple graph and semi-supervision techniques have been successfully introduced into the nonnegative matrix factorization (NMF) model for taking full advantage of the manifold structure and priori information of data to capture excellent low-dimensional data representation. However, the existing methods do not consider the sparse constraint, which can enhance the local learning ability and improve the performance in practical applications. To overcome this limitation, a novel NMF-based data representation method, namely, the multiple graph adaptive regularized semi-supervised nonnegative matrix factorization with sparse constraint (MSNMFSC) is developed in this paper for obtaining the sparse and discriminative data representation and increasing the quality of decomposition of NMF. Particularly, based on the standard NMF, the proposed MSNMFSC method combines the multiple graph adaptive regularization, the limited supervised information and the sparse constraint together to learn the more discriminative parts-based data representation. Moreover, the convergence analysis of MSNMFSC is studied. Experiments are conducted on several practical image datasets in clustering tasks, and the clustering results have shown that MSNMFSC achieves better performance than several most related NMF-based methods.

Deconvolution algorithms for inference of the cell-type composition of the spatial transcriptome

The spatial transcriptome has enabled researchers to resolve transcriptome expression profiles while preserving information about cell location to better understand the complex biological processes that occur in organisms. Due to technical limitations, the current high-throughput spatial transcriptome sequencing methods (known as next-generation sequencing with spatial barcoding methods or spot-based methods) cannot achieve single-cell resolution. A single measurement site, called a spot, in these technologies frequently contains multiple cells of various types. Computational tools for determining the cellular composition of a spot have emerged as a way to break through these limitations. These tools are known as deconvolution tools. Recently, a couple of deconvolution tools based on different strategies have been developed and have shown promise in different aspects. The resulting single-cell resolution expression profiles and/or single-cell composition of spots will significantly affect downstream data mining; thus, it is crucial to choose a suitable deconvolution tool. In this review, we present a list of currently available tools for spatial transcriptome deconvolution, categorize them based on the strategies they employ, and explain their advantages and limitations in detail in order to guide the selection of these tools in future studies.

PBRM1 presents a potential prognostic marker and therapeutic target in duodenal papillary carcinoma

BACKGROUND

Due to its rarity, duodenal papillary carcinoma (DPC) is seldom studied as a unique disease and no specific molecular features or treatment guidelines are provided.

METHODS

Whole-exome sequencing was performed to gain new insights into the DPC mutation landscape and to identify potential signalling pathways and therapeutic targets. Mechanistically, immunohistochemistry (IHC), immunofluorescence, RNA-seq, ATAC-seq and in vitro cell function experiments were performed to confirm the underlying mechanisms.

RESULTS

We described the mutational landscape of DPC for the first time as a group of rare tumours with a high frequency of dysregulation in the chromatin remodelling pathway, particularly PBRM1-inactivating mutations that are significantly higher than duodenal adenocarcinomas and ampullary adenocarcinoma (27% vs. 0% vs. 7%, p < .01). In vitro cell experiments showed that downregulation of PBRM1 expression could significantly promote the cancer progression and epithelial-to-mesenchymal transition via the PBRM1-c-JUN-VIM axis. The IHC data indicated that PBRM1 deficiency (p = .047) and c-JUN expression (p < .001) were significantly associated with poor prognosis. Meanwhile, the downregulation of PBRM1 expression in HUTU-80 cells was sensitive to radiation, which may be due to the suppression of c-JUN by irradiation.

CONCLUSIONS

Our findings define a novel molecular subgroup of PBRM1-inactivating mutations in DPC. PBRM1 play an important role in DPC progression and may serve as a potential therapeutic target and prognostic indicator.

Identification of Functional Microbial Modules Through Network-Based Analysis of Meta-Microbial Features Using Matrix Factorization

As the microbiome is composed of a variety of microbial interactions, it is imperative in microbiome research to identify a microbial sub-community that collectively conducts a specific function. However, current methodologies have been highly limited to analyzing conditional abundance changes of individual microorganisms without considering group-wise collective microbial features. To overcome this limitation, we developed a network-based method using nonnegative matrix factorization (NMF) to identify functional meta-microbial features (MMFs) that, as a group, better discriminate specific environmental conditions of samples using microbiome data. As proof of concept, large-scale human microbiome data collected from different body sites were used to identify body site-specific MMFs by applying NMF. The statistical test for MMFs led us to identify highly discriminative MMFs on sample classes, called synergistic MMFs (SYMMFs). Finally, we constructed a SYMMF-based microbial interaction network (SYMMF-net) by integrating all of the SYMMF information. Network analysis revealed core microbial modules closely related to critical sample properties. Similar results were also found when the method was applied to various disease-associated microbiome data. The developed method interprets high-dimensional microbiome data by identifying functional microbial modules on sample properties and intuitively representing their systematic relationships via a microbial network.

Learning discriminative and structural samples for rare cell types with deep generative model

Cell types (subpopulations) serve as bio-markers for the diagnosis and therapy of complex diseases, and single-cell RNA-sequencing (scRNA-seq) measures expression of genes at cell level, paving the way for the identification of cell types. Although great efforts have been devoted to this issue, it remains challenging to identify rare cell types in scRNA-seq data because of the few-shot problem, lack of interpretability and separation of generating samples and clustering of cells. To attack these issues, a novel deep generative model for leveraging the small samples of cells (aka scLDS2) is proposed by precisely estimating the distribution of different cells, which discriminate the rare and non-rare cell types with adversarial learning. Specifically, to enhance interpretability of samples, scLDS2 generates the sparse faked samples of cells with $\ell _1$-norm, where the relations among cells are learned, facilitating the identification of cell types. Furthermore, scLDS2 directly obtains cell types from the generated samples by learning the block structure such that cells belonging to the same types are similar to each other with the nuclear-norm. scLDS2 joins the generation of samples, classification of the generated and truth samples for cells and feature extraction into a unified generative framework, which transforms the rare cell types detection problem into a classification problem, paving the way for the identification of cell types with joint learning. The experimental results on 20 datasets demonstrate that scLDS2 significantly outperforms 17 state-of-the-art methods in terms of various measurements with 25.12% improvement in adjusted rand index on average, providing an effective strategy for scRNA-seq data with rare cell types. (The software is coded using python, and is freely available for academic https://github.com/xkmaxidian/scLDS2).

Knowledge-graph-based cell-cell communication inference for spatially resolved transcriptomic data with SpaTalk

Spatially resolved transcriptomics provides genetic information in space toward elucidation of the spatial architecture in intact organs and the spatially resolved cell-cell communications mediating tissue homeostasis, development, and disease. To facilitate inference of spatially resolved cell-cell communications, we here present SpaTalk, which relies on a graph network and knowledge graph to model and score the ligand-receptor-target signaling network between spatially proximal cells by dissecting cell-type composition through a non-negative linear model and spatial mapping between single-cell transcriptomic and spatially resolved transcriptomic data. The benchmarked performance of SpaTalk on public single-cell spatial transcriptomic datasets is superior to that of existing inference methods. Then we apply SpaTalk to STARmap, Slide-seq, and 10X Visium data, revealing the in-depth communicative mechanisms underlying normal and disease tissues with spatial structure. SpaTalk can uncover spatially resolved cell-cell communications for single-cell and spot-based spatially resolved transcriptomic data universally, providing valuable insights into spatial inter-cellular tissue dynamics.

Cell-cell communication is a vital feature involving numerous biological processes. Here, the authors develop SpaTalk, a cell-cell communication inference method using knowledge graph for spatially resolved transcriptomic data, providing valuable insights into spatial intercellular tissue dynamics.

Integrative pharmacogenomics revealed three subtypes with different immune landscapes and specific therapeutic responses in lung adenocarcinoma

Background

Pharmacogenomics is crucial for individualized drug therapy and plays an increasingly vital role in precision medicine decision-making. However, pharmacogenomics-based molecular subtypes and their potential clinical significance remain primarily unexplored in lung adenocarcinoma (LUAD).

Methods

A total of 2065 samples were recruited from eight independent cohorts. Pharmacogenomics data were generated from the profiling of relative inhibition simultaneously in mixtures (PRISM) and the genomics of drug sensitivity in cancer (GDSC) databases. Multiple bioinformatics approaches were performed to identify pharmacogenomics-based subtypes and find subtype-specific properties.

Results

Three reproducible molecular subtypes were found, which were independent prognostic factors and highly associated with stage, survival status, and accepted molecular subtypes. Pharmacogenomics-based subtypes had distinct molecular characteristics: S-Ⅰ was inflammatory, proliferative, and immune-evasion; S-Ⅱ was proliferative and genetics-driven; S-III was metabolic and methylation-driven. Finally, our study provided subtype-guided personalized treatment strategies: Immune checkpoint blockers (ICBs), doxorubicin, tipifarnib, AZ628, and AZD6244 were for S-Ⅰ; Cisplatin, camptothecin, roscovitine, and A.443654 were for S-Ⅱ; Docetaxel, paclitaxel, vinorelbine, and BIBW2992 were for S-III.

Conclusion

We provided a novel molecular classification strategy and revealed three pharmacogenomics-based subtypes for LUAD patients, which uncovered potential subtype-related and patient-specific therapeutic strategies.

Cell-lineage controlled epigenetic regulation in glioblastoma stem cells determines functionally distinct subgroups and predicts patient survival

There is ample support for developmental regulation of glioblastoma stem cells. To examine how cell lineage controls glioblastoma stem cell function, we present a cross-species epigenome analysis of mouse and human glioblastoma stem cells. We analyze and compare the chromatin-accessibility landscape of nine mouse glioblastoma stem cell cultures of three defined origins and 60 patient-derived glioblastoma stem cell cultures by assay for transposase-accessible chromatin using sequencing. This separates the mouse cultures according to cell of origin and identifies three human glioblastoma stem cell clusters that show overlapping characteristics with each of the mouse groups, and a distribution along an axis of proneural to mesenchymal phenotypes. The epigenetic-based human glioblastoma stem cell clusters display distinct functional properties and can separate patient survival. Cross-species analyses reveals conserved epigenetic regulation of mouse and human glioblastoma stem cells. We conclude that epigenetic control of glioblastoma stem cells primarily is dictated by developmental origin which impacts clinically relevant glioblastoma stem cell properties and patient survival.

The epigenetic regulation of glioblastoma stem cell (GSC) function remains poorly understood. Here, the authors compare the chromatin accessibility landscape of GSC cultures from mice and patients and suggest that the epigenome of GSCs is cell lineage-regulated and could predict patient survival.

DNA Methylation Profiles of Ovarian Clear Cell Carcinoma

BACKGROUND

Ovarian clear cell carcinoma (OCCC) is a rare ovarian cancer histotype that tends to be resistant to standard platinum-based chemotherapeutics. We sought to better understand the role of DNA methylation in clinical and biological subclassification of OCCC.

METHODS

We interrogated genome-wide methylation using DNA from fresh frozen tumors from 271 cases, applied nonsmooth nonnegative matrix factorization (nsNMF) clustering, and evaluated clinical associations and biological pathways.

RESULTS

Two approximately equally sized clusters that associated with several clinical features were identified. Compared with Cluster 2 (N = 137), Cluster 1 cases (N = 134) presented at a more advanced stage, were less likely to be of Asian ancestry, and tended to have poorer outcomes including macroscopic residual disease following primary debulking surgery (P < 0.10). Subset analyses of targeted tumor sequencing and IHC data revealed that Cluster 1 tumors showed TP53 mutation and abnormal p53 expression, and Cluster 2 tumors showed aneuploidy and ARID1A/PIK3CA mutation (P < 0.05). Cluster-defining CpGs included 1,388 CpGs residing within 200 bp of the transcription start sites of 977 genes; 38% of these genes (N = 369 genes) were differentially expressed across cluster in transcriptomic subset analysis (P < 10-4). Differentially expressed genes were enriched for six immune-related pathways, including IFNα and IFNγ responses (P < 10-6).

CONCLUSIONS

DNA methylation clusters in OCCC correlate with disease features and gene expression patterns among immune pathways.

IMPACT

This work serves as a foundation for integrative analyses that better understand the complex biology of OCCC in an effort to improve potential for development of targeted therapeutics.

Evolving Transcriptomic Profiles From Single-Cell RNA-Seq Data Using Nature-Inspired Multiobjective Optimization

Transcriptomic profiling plays an important role in post-genomic analysis. Especially, the single-cell RNA-seq technology has advanced our understanding of gene expression from cell population level into individual cell level. Many computational methods have been proposed to decipher transcriptomic profiles from those RNA-seq data. However, most of the related algorithms suffer from realistic restrictions such as high dimensionality and premature convergence. In this paper, we propose and formulate an evolutionary multiobjective blind compressed sensing (EMOBCS) to address those problems for evolving transcriptomic profiles from single-cell RNA-seq data. In the proposed framework, to characterize various gene expression profile models, two objective functions including chi-squared kernel score and euclidean distance of different gene expression profiles are formulated. After that, multiobjective blind compressed sensing based on artificial bee colony is designed to optimize the two objective functions on single-cell RNA-seq data by proposing a rank probability model and two new search strategies into the cooperative convolution framework in an unbiased manner. To demonstrate its effectiveness, extensive experiments have been conducted, comparing the proposed algorithm with 14 algorithms including eight state-of-the-art algorithms and six different EMOBCS algorithms under different search strategies on 10 single-cell RNA-seq datasets and one case study. The experimental results reveal that the proposed algorithm is better than or comparable with those compared algorithms. Furthermore, we also conduct the time complexity analysis, convergence analysis, and parameter analysis to demonstrate various properties of EMOBCS.

Mitral Valve Segmentation Using Robust Nonnegative Matrix Factorization

Analyzing and understanding the movement of the mitral valve is of vital importance in cardiology, as the treatment and prevention of several serious heart diseases depend on it. Unfortunately, large amounts of noise as well as a highly varying image quality make the automatic tracking and segmentation of the mitral valve in two-dimensional echocardiographic videos challenging. In this paper, we present a fully automatic and unsupervised method for segmentation of the mitral valve in two-dimensional echocardiographic videos, independently of the echocardiographic view. We propose a bias-free variant of the robust non-negative matrix factorization (RNMF) along with a window-based localization approach, that is able to identify the mitral valve in several challenging situations. We improve the average f1-score on our dataset of 10 echocardiographic videos by 0.18 to a f1-score of 0.56.

Population sequencing data reveal a compendium of mutational processes in the human germ line

Biological mechanisms underlying human germline mutations remain largely unknown. We statistically decompose variation in the rate and spectra of mutations along the genome using volume-regularized nonnegative matrix factorization. The analysis of a sequencing dataset (TOPMed) reveals nine processes that explain the variation in mutation properties between loci. We provide a biological interpretation for seven of these processes. We associate one process with bulky DNA lesions that are resolved asymmetrically with respect to transcription and replication. Two processes track direction of replication fork and replication timing, respectively. We identify a mutagenic effect of active demethylation primarily acting in regulatory regions and a mutagenic effect of long interspersed nuclear elements. We localize a mutagenic process specific to oocytes from population sequencing data. This process appears transcriptionally asymmetric.

Supervised application of internal validation measures to benchmark dimensionality reduction methods in scRNA-seq data

A typical single-cell RNA sequencing (scRNA-seq) experiment will measure on the order of 20 000 transcripts and thousands, if not millions, of cells. The high dimensionality of such data presents serious complications for traditional data analysis methods and, as such, methods to reduce dimensionality play an integral role in many analysis pipelines. However, few studies have benchmarked the performance of these methods on scRNA-seq data, with existing comparisons assessing performance via downstream analysis accuracy measures, which may confound the interpretation of their results. Here, we present the most comprehensive benchmark of dimensionality reduction methods in scRNA-seq data to date, utilizing over 300 000 compute hours to assess the performance of over 25 000 low-dimension embeddings across 33 dimensionality reduction methods and 55 scRNA-seq datasets. We employ a simple, yet novel, approach, which does not rely on the results of downstream analyses. Internal validation measures (IVMs), traditionally used as an unsupervised method to assess clustering performance, are repurposed to measure how well-formed biological clusters are after dimensionality reduction. Performance was further evaluated over nearly 200 000 000 iterations of DBSCAN, a density-based clustering algorithm, showing that hyperparameter optimization using IVMs as the objective function leads to near-optimal clustering. Methods were also assessed on the extent to which they preserve the global structure of the data, and on their computational memory and time requirements across a large range of sample sizes. Our comprehensive benchmarking analysis provides a valuable resource for researchers and aims to guide best practice for dimensionality reduction in scRNA-seq analyses, and we highlight Latent Dirichlet Allocation and Potential of Heat-diffusion for Affinity-based Transition Embedding as high-performing algorithms.

De novo mutational signature discovery in tumor genomes using SparseSignatures

Cancer is the result of mutagenic processes that can be inferred from tumor genomes by analyzing rate spectra of point mutations, or "mutational signatures". Here we present SparseSignatures, a novel framework to extract signatures from somatic point mutation data. Our approach incorporates a user-specified background signature, employs regularization to reduce noise in non-background signatures, uses cross-validation to identify the number of signatures, and is scalable to large datasets. We show that SparseSignatures outperforms current state-of-the-art methods on simulated data using a variety of standard metrics. We then apply SparseSignatures to whole genome sequences of pancreatic and breast tumors, discovering well-differentiated signatures that are linked to known mutagenic mechanisms and are strongly associated with patient clinical features.

Deep Non-Negative Matrix Factorization Architecture Based on Underlying Basis Images Learning

The non-negative matrix factorization (NMF) algorithm represents the original image as a linear combination of a set of basis images. This image representation method is in line with the idea of "parts constitute a whole" in human thinking. The existing deep NMF performs deep factorization on the coefficient matrix. In these methods, the basis images used to represent the original image is essentially obtained by factorizing the original images once. To extract features reflecting the deep localization characteristics of images, a novel deep NMF architecture based on underlying basis images learning is proposed for the first time. The architecture learns the underlying basis images by deep factorization on the basis images matrix. The deep factorization architecture proposed in this paper has strong interpretability. To implement this architecture, this paper proposes a deep non-negative basis matrix factorization algorithm to obtain the underlying basis images. Then, the objective function is established with an added regularization term, which directly constrains the basis images matrix to obtain the basis images with good local characteristics, and a regularized deep non-negative basis matrix factorization algorithm is proposed. The regularized deep nonlinear non-negative basis matrix factorization algorithm is also proposed to handle pattern recognition tasks with complex data. This paper also theoretically proves the convergence of the algorithm. Finally, the experimental results show that the deep NMF architecture based on the underlying basis images learning proposed in this paper can obtain better recognition performance than the other state-of-the-art methods.

Uniform Distribution Non-Negative Matrix Factorization for Multiview Clustering

Multiview data processing has attracted sustained attention as it can provide more information for clustering. To integrate this information, one often utilizes the non-negative matrix factorization (NMF) scheme which can reduce the data from different views into the subspace with the same dimension. Motivated by the clustering performance being affected by the distribution of the data in the learned subspace, a tri-factorization-based NMF model with an embedding matrix is proposed in this article. This model tends to generate decompositions with uniform distribution, such that the learned representations are more discriminative. As a result, the obtained consensus matrix can be a better representative of the multiview data in the subspace, leading to higher clustering performance. Also, a new lemma is proposed to provide the formulas about the partial derivation of the trace function with respect to an inner matrix, together with its theoretical proof. Based on this lemma, a gradient-based algorithm is developed to solve the proposed model, and its convergence and computational complexity are analyzed. Experiments on six real-world datasets are performed to show the advantages of the proposed algorithm, with comparison to the existing baseline methods.

Feature Extraction Using Sparse Kernel Non-Negative Matrix Factorization for Rolling Element Bearing Diagnosis

For early fault detection of a bearing, the localized defect generally brings a complex vibration signal, so it is difficult to detect the periodic transient characteristics from the signal spectrum using conventional bearing fault diagnosis methods. Therefore, many matrix analysis technologies, such as singular value decomposition (SVD) and reweighted SVD (RSVD), were proposed recently to solve this problem. However, such technologies also face failure in bearing fault detection due to the poor interpretability of the obtained eigenvector. Non-negative Matrix Factorization (NMF), as a part-based representation algorithm, can extract low-rank basis spaces with natural sparsity from the time–frequency representation. It performs excellent interpretability of the factor matrices due to its non-negative constraints. By this virtue, NMF can extract the fault feature by separating the frequency bands of resonance regions from the amplitude spectrogram automatically. In this paper, a new feature extraction method based on sparse kernel NMF (KNMF) was proposed to extract the fault features from the amplitude spectrogram in greater depth. By decomposing the amplitude spectrogram using the kernel-based NMF model with L1 regularization, sparser spectral bases can be obtained. Using KNMF with the linear kernel function, the time–frequency distribution of the vibration signal can be decomposed into a subspace with different frequency bands. Thus, we can extract the fault features, a series of periodic impulses, from the decomposed subspace according to the sparse frequency bands in the spectral bases. As a result, the proposed method shows a very high performance in extracting fault features, which is verified by experimental investigations and benchmarked by the Fast Kurtogram, SVD and NMF-based methods.

SPOTlight: seeded NMF regression to deconvolute spatial transcriptomics spots with single-cell transcriptomes

Spatially resolved gene expression profiles are key to understand tissue organization and function. However, spatial transcriptomics (ST) profiling techniques lack single-cell resolution and require a combination with single-cell RNA sequencing (scRNA-seq) information to deconvolute the spatially indexed datasets. Leveraging the strengths of both data types, we developed SPOTlight, a computational tool that enables the integration of ST with scRNA-seq data to infer the location of cell types and states within a complex tissue. SPOTlight is centered around a seeded non-negative matrix factorization (NMF) regression, initialized using cell-type marker genes and non-negative least squares (NNLS) to subsequently deconvolute ST capture locations (spots). Simulating varying reference quantities and qualities, we confirmed high prediction accuracy also with shallowly sequenced or small-sized scRNA-seq reference datasets. SPOTlight deconvolution of the mouse brain correctly mapped subtle neuronal cell states of the cortical layers and the defined architecture of the hippocampus. In human pancreatic cancer, we successfully segmented patient sections and further fine-mapped normal and neoplastic cell states. Trained on an external single-cell pancreatic tumor references, we further charted the localization of clinical-relevant and tumor-specific immune cell states, an illustrative example of its flexible application spectrum and future potential in digital pathology.

Comparison of sparse biclustering algorithms for gene expression datasets

MOTIVATION

Gene clustering and sample clustering are commonly used to find patterns in gene expression datasets. However, genes may cluster differently in heterogeneous samples (e.g. different tissues or disease states), whilst traditional methods assume that clusters are consistent across samples. Biclustering algorithms aim to solve this issue by performing sample clustering and gene clustering simultaneously. Existing reviews of biclustering algorithms have yet to include a number of more recent algorithms and have based comparisons on simplistic simulated datasets without specific evaluation of biclusters in real datasets, using less robust metrics.

RESULTS

We compared four classes of sparse biclustering algorithms on a range of simulated and real datasets. All algorithms generally struggled on simulated datasets with a large number of genes or implanted biclusters. We found that Bayesian algorithms with strict sparsity constraints had high accuracy on the simulated datasets and did not require any post-processing, but were considerably slower than other algorithm classes. We found that non-negative matrix factorisation algorithms performed poorly, but could be re-purposed for biclustering through a sparsity-inducing post-processing procedure we introduce; one such algorithm was one of the most highly ranked on real datasets. In a multi-tissue knockout mouse RNA-seq dataset, the algorithms rarely returned clusters containing samples from multiple different tissues, whilst such clusters were identified in a human dataset of more closely related cell types (sorted blood cell subsets). This highlights the need for further thought in the design and analysis of multi-tissue studies to avoid differences between tissues dominating the analysis.

AVAILABILITY

Code to run the analysis is available at https://github.com/nichollskc/biclust_comp, including wrappers for each algorithm, implementations of evaluation metrics, and code to simulate datasets and perform pre- and post-processing. The full tables of results are available at https://doi.org/10.5281/zenodo.4581206.

Glioblastomas acquire myeloid-affiliated transcriptional programs via epigenetic immunoediting to elicit immune evasion

Glioblastoma multiforme (GBM) is an aggressive brain tumor for which current immunotherapy approaches have been unsuccessful. Here, we explore the mechanisms underlying immune evasion in GBM. By serially transplanting GBM stem cells (GSCs) into immunocompetent hosts, we uncover an acquired capability of GSCs to escape immune clearance by establishing an enhanced immunosuppressive tumor microenvironment. Mechanistically, this is not elicited via genetic selection of tumor subclones, but through an epigenetic immunoediting process wherein stable transcriptional and epigenetic changes in GSCs are enforced following immune attack. These changes launch a myeloid-affiliated transcriptional program, which leads to increased recruitment of tumor-associated macrophages. Furthermore, we identify similar epigenetic and transcriptional signatures in human mesenchymal subtype GSCs. We conclude that epigenetic immunoediting may drive an acquired immune evasion program in the most aggressive mesenchymal GBM subtype by reshaping the tumor immune microenvironment.

Gene Expression Classifier Reveals Prognostic Osteosarcoma Microenvironment Molecular Subtypes

Osteosarcoma (OSA) is the most common bone malignancy and displays high heterogeneity of molecular phenotypes. This study aimed to characterize the molecular features of OSA by developing a classification system based on the gene expression profile of the tumor microenvironment. Integrative analysis was performed using specimens and clinical information for OSA patients from the TARGET program. Using a matrix factorization method, we identified two molecular subtypes significantly associated with prognosis, S1 (infiltration type) and S2 (escape type). Both subtypes displayed unique features of functional significance features and cellular infiltration characteristics. We determined that immune and stromal infiltrates were abundant in subtype S1 compare to that in subtype S2. Furthermore, higher expression of immune checkpoint PDCD1LG2 and HAVCR2 was associated with improved prognosis, while a preferable chemotherapeutic response was associated with FAP-positive fibroblasts in subtype S1. Alternatively, subtype S2 is characterized by a lack of effective cytotoxic responses and loss of major histocompatibility complex class I molecule expression. A gene classifier was ultimately generated to enable OSA classification and the results were confirmed using the GSE21257 validation set. Correlations between the percentage of fibroblasts and/or fibrosis and CD8⁺ cells, and their clinical responses to chemotherapy were assessed and verified based on 47 OSA primary tumors. This study established a new OSA classification system for stratifying OSA patient risk, thereby further defining the genetic diversity of OSA and allowing for improved efficiency of personalized therapy.

Target Area Extraction Algorithm for the In Vivo Fluorescence Imaging of Small Animals

Bio-optical imaging can noninvasively describe specific biochemical reaction events in small animals using endogenous or exogenous imaging reagents to label cells, proteins, or DNA. The fluorescence optical bio-imaging system excites the fluorescent group to a high energy state by excitation light and then generates emission light. However, many substances in the organism will also emit fluorescence after being excited by the excitation light, and the nonspecific fluorescence generated will affect the detection sensitivity. This paper designs and develops a set of high-level biosafety in vivo fluorescence imaging system for small animals suitable for virology research and proposes a target area extraction algorithm for fluorescence images. The fluorescence image target extraction algorithm first maps the nonlinear separation data in the low-dimensional space to the high-dimensional space. Then, based on the analysis of the characteristics of the fluorescent region, a method for discriminating the target fluorescent region based on the two-step entropy function is proposed, and the real target fluorescent region is obtained according to the set connected region. Based on the experiment of collecting and analyzing the in vivo fluorescent images of mice, it is verified that the proposed algorithm can automatically extract the target fluorescent region better than the classical linear model. It shows that the proposed algorithm is less affected by background fluorescence, and the estimated separated spectrum based on this method is closer to the real target spectrum.

A practical framework and online tool for mutational signature analyses show inter-tissue variation and driver dependencies

Mutational signatures are patterns of mutations that arise during tumorigenesis. We present an enhanced, practical framework for mutational signature analyses. Applying these methods on 3,107 whole genome sequenced (WGS) primary cancers of 21 organs reveals known signatures and nine previously undescribed rearrangement signatures. We highlight inter-organ variability of signatures and present a way of visualizing that diversity, reinforcing our findings in an independent analysis of 3,096 WGS metastatic cancers. Signatures with a high level of genomic instability are dependent on TP53 dysregulation. We illustrate how uncertainty in mutational signature identification and assignment to samples affects tumor classification, reinforcing that using multiple orthogonal mutational signature data is not only beneficial, it is essential for accurate tumor stratification. Finally, we present a reference web-based tool for cancer and experimentally-generated mutational signatures, called Signal (https://signal.mutationalsignatures.com), that also supports performing mutational signature analyses.

Exploring background mutational processes to decipher cancer genetic heterogeneity

Much remains unknown about the progression and heterogeneity of mutational processes in different cancers and their diagnostic and clinical potential. A growing body of evidence supports mutation rate dependence on the local DNA sequence context for various types of mutations. We propose several tools for the analysis of cancer context-dependent mutations, which are implemented in an online computational framework MutaGene. The framework explores DNA context-dependent mutational patterns and underlying somatic cancer mutagenesis, analyzes mutational profiles of cancer samples, identifies the combinations of underlying mutagenic processes including those related to infidelity of DNA replication and repair machinery, and various other endogenous and exogenous mutagenic factors. As a result, the combination of mutagenic processes can be identified in any query sample with subsequent comparison to mutational profiles derived from malignant and benign samples. In addition, mutagen or cancer-specific mutational background models are applied to calculate expected DNA and protein site mutability to decouple relative contributions of mutagenesis and selection in carcinogenesis, thus elucidating the site-specific driving events in cancer. MutaGene is freely available at https://www.ncbi.nlm.nih.gov/projects/mutagene/.

Molecular signatures of X chromosome inactivation and associations with clinical outcomes in epithelial ovarian cancer

X chromosome inactivation (XCI) is a key epigenetic gene expression regulatory process, which may play a role in women's cancer. In particular tissues, some genes are known to escape XCI, yet patterns of XCI in ovarian cancer (OC) and their clinical associations are largely unknown. To examine XCI in OC, we integrated germline genotype with tumor copy number, gene expression and DNA methylation information from 99 OC patients. Approximately 10% of genes showed different XCI status (either escaping or being subject to XCI) compared with the studies of other tissues. Many of these genes are known oncogenes or tumor suppressors (e.g. DDX3X, TRAPPC2 and TCEANC). We also observed strong association between cis promoter DNA methylation and allele-specific expression imbalance (P = 2.0 × 10-10). Cluster analyses of the integrated data identified two molecular subgroups of OC patients representing those with regulated (N = 47) and dysregulated (N = 52) XCI. This XCI cluster membership was associated with expression of X inactive specific transcript (P = 0.002), a known driver of XCI, as well as age, grade, stage, tumor histology and extent of residual disease following surgical debulking. Patients with dysregulated XCI (N = 52) had shorter time to recurrence (HR = 2.34, P = 0.001) and overall survival time (HR = 1.87, P = 0.02) than those with regulated XCI, although results were attenuated after covariate adjustment. Similar findings were observed when restricted to high-grade serous tumors. We found evidence of a unique OC XCI profile, suggesting that XCI may play an important role in OC biology. Additional studies to examine somatic changes with paired tumor-normal tissue are needed.

Proteomics identifies new therapeutic targets of early-stage hepatocellular carcinoma

Hepatocellular carcinoma is the third leading cause of deaths from cancer worldwide. Infection with the hepatitis B virus is one of the leading risk factors for developing hepatocellular carcinoma, particularly in East Asia¹. Although surgical treatment may be effective in the early stages, the five-year overall rate of survival after developing this cancer is only 50-70%². Here, using proteomic and phospho-proteomic profiling, we characterize 110 paired tumour and non-tumour tissues of clinical early-stage hepatocellular carcinoma related to hepatitis B virus infection. Our quantitative proteomic data highlight heterogeneity in early-stage hepatocellular carcinoma: we used this to stratify the cohort into the subtypes S-I, S-II and S-III, each of which has a different clinical outcome. S-III, which is characterized by disrupted cholesterol homeostasis, is associated with the lowest overall rate of survival and the greatest risk of a poor prognosis after first-line surgery. The knockdown of sterol O-acyltransferase 1 (SOAT1)-high expression of which is a signature specific to the S-III subtype-alters the distribution of cellular cholesterol, and effectively suppresses the proliferation and migration of hepatocellular carcinoma. Finally, on the basis of a patient-derived tumour xenograft mouse model of hepatocellular carcinoma, we found that treatment with avasimibe, an inhibitor of SOAT1, markedly reduced the size of tumours that had high levels of SOAT1 expression. The proteomic stratification of early-stage hepatocellular carcinoma presented in this study provides insight into the tumour biology of this cancer, and suggests opportunities for personalized therapies that target it.

A framework for intelligent Twitter data analysis with non-negative matrix factorization

Purpose

The purpose of this paper is to propose a framework for intelligent analysis of Twitter data. The purpose of the framework is to allow users to explore a collection of tweets by extracting topics with semantic relevance. In this way, it is possible to detect groups of tweets related to new technologies, events and other topics that are automatically discovered.

Design/methodology/approach

The framework is based on a three-stage process. The first stage is devoted to dataset creation by transforming a collection of tweets in a dataset according to the vector space model. The second stage, which is the core of the framework, is centered on the use of non-negative matrix factorizations (NMF) for extracting human-interpretable topics from tweets that are eventually clustered. The number of topics can be user-defined or can be discovered automatically by applying subtractive clustering as a preliminary step before factorization. Cluster analysis and word-cloud visualization are used in the last stage to enable intelligent data analysis.

Findings

The authors applied the framework to a case study of three collections of Italian tweets both with manual and automatic selection of the number of topics. Given the high sparsity of Twitter data, the authors also investigated the influence of different initializations mechanisms for NMF on the factorization results. Numerical comparisons confirm that NMF could be used for clustering as it is comparable to classical clustering techniques such as spherical k-means. Visual inspection of the word-clouds allowed a qualitative assessment of the results that confirmed the expected outcomes.

Originality/value

The proposed framework enables a collaborative approach between users and computers for an intelligent analysis of Twitter data. Users are faced with interpretable descriptions of tweet clusters, which can be interactively refined with few adjustable parameters. The resulting clusters can be used for intelligent selection of tweets, as well as for further analytics concerning the impact of products, events, etc. in the social network.