Integrating Multi-omics Data for Alzheimer's Disease to Explore Its Biomarkers Via the Hypergraph-Regularized Joint Deep Semi-Non-Negative Matrix Factorization Algorithm

Alzheimer's disease (AD) is a progressive and irreversible neurodegenerative disorder. Its etiology may be associated with genetic, environmental, and lifestyle factors. With the advancement of technology, the integration of genomics, transcriptomics, and imaging data related to AD allows simultaneous exploration of molecular information at different levels and their interaction within the organism. This paper proposes a hypergraph-regularized joint deep semi-non-negative matrix factorization (HR-JDSNMF) algorithm to integrate positron emission tomography (PET), single-nucleotide polymorphism (SNP), and gene expression data for AD. The method employs matrix factorization techniques to nonlinearly decompose the original data at multiple layers, extracting deep features from different omics data, and utilizes hypergraph mining to uncover high-order correlations among the three types of data. Experimental results demonstrate that this approach outperforms several matrix factorization-based algorithms and effectively identifies multi-omics biomarkers for AD. Additionally, single-cell RNA sequencing (scRNA-seq) data for AD were collected, and genes within significant modules were used to categorize different types of cell clusters into high and low-risk cell groups. Finally, the study extensively explores the differences in differentiation and communication between these two cell types. The multi-omics biomarkers unearthed in this study can serve as valuable references for the clinical diagnosis and drug target discovery for AD. The realization of the algorithm in this paper code is available at https://github.com/ShubingKong/HR-JDSNMF .

Spinal cord and brain atrophy patterns in neuromyelitis optica spectrum disorder and multiple sclerosis

BACKGROUND

Spinal cord and brain atrophy are common in neuromyelitis optica spectrum disorder (NMOSD) and relapsing-remitting multiple sclerosis (RRMS) but harbor distinct patterns accounting for disability and cognitive impairment.

METHODS

This study included 209 NMOSD and 304 RRMS patients and 436 healthy controls. Non-negative matrix factorization was used to parse differences in spinal cord and brain atrophy at subject level into distinct patterns based on structural MRI. The weights of patterns were obtained using a linear regression model and associated with Expanded Disability Status Scale (EDSS) and cognitive scores. Additionally, patients were divided into cognitive impairment (CI) and cognitive preservation (CP) groups.

RESULTS

Three patterns were observed in NMOSD: (1) Spinal Cord-Deep Grey Matter (SC-DGM) pattern was associated with high EDSS scores and decline of visuospatial memory function; (2) Frontal-Temporal pattern was associated with decline of language learning function; and (3) Cerebellum-Brainstem pattern had no observed association. Patients with CI had higher weights of SC-DGM pattern than CP group. Three patterns were observed in RRMS: (1) DGM pattern was associated with high EDSS scores, decreased information processing speed, and decreased language learning and visuospatial memory functions; (2) Frontal-Temporal pattern was associated with overall cognitive decline; and (3) Occipital pattern had no observed association. Patients with CI trended to have higher weights of DGM and Frontal-Temporal patterns than CP group.

CONCLUSION

This study estimated the heterogeneity of spinal cord and brain atrophy patterns in NMOSD and RRMS patients at individual level, and evaluated the clinical relevance of these patterns, which may contribute to stratifying participants for targeted therapy.

Singular Value Decomposition-Driven Non-negative Matrix Factorization with Application to Identify the Association Patterns of Sarcoma Recurrence

Sarcomas are malignant tumors from mesenchymal tissue and are characterized by their complexity and diversity. The high recurrence rate making it important to understand the mechanisms behind their recurrence and to develop personalized treatments and drugs. However, previous studies on the association patterns of multi-modal data on sarcoma recurrence have overlooked the fact that genes do not act independently, but rather function within signaling pathways. Therefore, this study collected 290 whole solid images, 869 gene and 1387 pathway data of over 260 sarcoma samples from UCSC and TCGA to identify the association patterns of gene-pathway-cell related to sarcoma recurrences. Meanwhile, considering that most multi-modal data fusion methods based on the joint non-negative matrix factorization (NMF) model led to poor experimental repeatability due to random initialization of factorization parameters, the study proposed the singular value decomposition (SVD)-driven joint NMF model by applying the SVD method to calculate initialized weight and coefficient matrices to achieve the reproducibility of the results. The results of the experimental comparison indicated that the SVD algorithm enhances the performance of the joint NMF algorithm. Furthermore, the representative module indicated a significant relationship between genes in pathways and image features. Multi-level analysis provided valuable insights into the connections between biological processes, cellular features, and sarcoma recurrence. In addition, potential biomarkers were uncovered, while various mechanisms of sarcoma recurrence were identified from an imaging genetic perspective. Overall, the SVD-NMF model affords a novel perspective on combining multi-omics data to explore the association related to sarcoma recurrence.

Abnormal interlimb coordination of motor developmental delay during infant crawling based on kinematic synergy analysis

BACKGROUND

Previous studies have reported that abnormal interlimb coordination is a typical characteristic of motor developmental delay (MDD) during human movement, which can be visually manifested as abnormal motor postures. Clinically, the scale assessments are usually used to evaluate interlimb coordination, but they rely heavily on the subjective judgements of therapists and lack quantitative analysis. In addition, although abnormal interlimb coordination of MDD have been studied, it is still unclear how this abnormality is manifested in physiology-related kinematic features.

OBJECTIVES

This study aimed to evaluate how abnormal interlimb coordination of MDD during infant crawling was manifested in the stability of joints and limbs, activation levels of synergies and intrasubject consistency from the kinematic synergies of tangential velocities of joints perspective.

METHODS

Tangential velocities of bilateral shoulder, elbow, wrist, hip, knee and ankle over time were computed from recorded three-dimensional joint trajectories in 40 infants with MDD [16 infants at risk of developmental delay, 11 infants at high risk of developmental delay, 13 infants with confirmed developmental delay (CDD group)] and 20 typically developing infants during hands-and-knees crawling. Kinematic synergies and corresponding activation coefficients were derived from those joint velocities using the non-negative matrix factorization algorithm. The variability accounted for yielded by those synergies and activation coefficients, and the synergy weightings in those synergies were used to measure the stability of joints and limbs. To quantify the activation levels of those synergies, the full width at half maximum and center of activity of activation coefficients were calculated. In addition, the intrasubject consistency was measured by the cosine similarity of those synergies and activation coefficients.

RESULTS

Interlimb coordination patterns during infant crawling were the combinations of four types of single-limb movements, which represent the dominance of each of the four limbs. MDD mainly reduced the stability of joints and limbs, and induced the abnormal activation levels of those synergies. Meanwhile, MDD generally reduced the intrasubject consistency, especially in CDD group.

CONCLUSIONS

These features have the potential for quantitatively evaluating abnormal interlimb coordination in assisting the clinical diagnosis and motor rehabilitation of MDD.

Understanding muscle coordination during gait based on muscle synergy and its association with symptoms in patients with knee osteoarthritis

OBJECTIVE

We aimed to investigate the muscle coordination differences between a control group and patients with mild and severe knee osteoarthritis (KOA) using muscle synergy analysis and determine whether muscle coordination was associated with symptoms of KOA.

METHOD

Fifty-three women with medial KOA and 19 control patients participated in the study. The gait analyses and muscle activity measurements of seven lower limb muscles were assessed using a motion capture system and electromyography. Gait speed and knee adduction moment impulse were calculated. The spatiotemporal components of muscle synergy were extracted using non-negative matrix factorization, and the dynamic motor control index during walking (walk-DMC) was computed. The number of muscle synergy and their spatiotemporal components were compared among the mild KOA, severe KOA, and control groups. Moreover, the association between KOA symptoms with walk-DMC and other gait parameters was evaluated using multi-linear regression analysis.

RESULTS

The number of muscle synergies was lower in mild and severe KOA compared with those in the control group. In synergy 1, the weightings of biceps femoris and gluteus medius in severe KOA were higher than that in the control group. In synergy 3, the weightings of higher tibial anterior and lower gastrocnemius lateralis were confirmed in the mild KOA group. Regression analysis showed that the walk-DMC was independently associated with knee-related symptoms of KOA after adjusting for the covariates.

CONCLUSIONS

Muscle coordination was altered in patients with KOA. The correlation between muscle coordination and KOA may be attributed to the knee-related symptoms. Key points • Patients with knee osteoarthritis (OA) experienced a deterioration in muscle coordination when walking. • Loss of muscle coordination was associated with severe knee-related symptoms in knee OA. • Considering muscle coordination as a knee OA symptom-related factor may provide improved treatment.

JLONMFSC: Clustering scRNA-seq data based on joint learning of non-negative matrix factorization and subspace clustering

The development of single cell RNA sequencing (scRNA-seq) has provided new perspectives to study biological problems at the single cell level. One of the key issues in scRNA-seq data analysis is to divide cells into several clusters for discovering the heterogeneity and diversity of cells. However, the existing scRNA-seq data are high-dimensional, sparse, and noisy, which challenges the existing single-cell clustering methods. In this study, we propose a joint learning framework (JLONMFSC) for clustering scRNA-seq data. In our method, the dimension of the original data is reduced to minimize the effect of noise. In addition, the graph regularized matrix factorization is used to learn the local features. Further, the Low-Rank Representation (LRR) subspace clustering is utilized to learn the global features. Finally, the joint learning of local features and global features is performed to obtain the results of clustering. We compare the proposed algorithm with eight state-of-the-art algorithms for clustering performance on six datasets, and the experimental results demonstrate that the JLONMFSC achieves better performance in all datasets. The code is avalable at https://github.com/lanbiolab/JLONMFSC.

Few sex-specific effects of fatigue on muscle synergies in a repetitive pointing task

Previous studies have identified some sex differences in how individual muscles change their activation during repetitive multi-joint arm motion-induced fatigue. However, little is known about how indicators of multi-muscle coordination change with fatigue in males and females. Fifty-six (29 females) asymptomatic young adults performed a repetitive, forward-backward pointing task until scoring 8/10 on a Borg CR10 scale while surface electromyographic activity of upper trapezius, anterior deltoid, biceps brachii, and triceps brachii was recorded. Activation coefficient, synergy structure, and relative weight of each muscle within synergies were calculated using the non-negative matrix factorization method. Two muscle synergies were extracted from the fatiguing task. The synergy structures were mostly preserved after fatigue, while the activation coefficients were altered. A significant Sex × Fatigue interaction effect showed more use of the anterior deltoid in males especially before fatigue in synergy 1 during shoulder stabilization (p = 0.04). As for synergy 2, it was characterized by variations in the relative weight of biceps, which was higher by 16 % in females compared to males (p = 0.04), and increased with fatigue (p = 0.03) during the elbow flexion acceleration phase and the deceleration phase of the backward pointing movement. Findings suggest that both sexes adapted to fatigue similarly, using fixed synergy structures, with alterations in synergy activation patterns and relative weights of individual muscles. Results support previous findings of an important role for the biceps and anterior deltoid in explaining sex differences in patterns of repetitive motion-induced upper limb fatigue.

A time-resolved meta-analysis of consensus gene expression profiles during human T-cell activation

BACKGROUND

The coordinated transcriptional regulation of activated T-cells is based on a complex dynamic behavior of signaling networks. Given an external stimulus, T-cell gene expression is characterized by impulse and sustained patterns over the course. Here, we analyze the temporal pattern of activation across different T-cell populations to develop consensus gene signatures for T-cell activation.

RESULTS

Here, we identify and verify general biomarker signatures robustly evaluating T-cell activation in a time-resolved manner. We identify time-resolved gene expression profiles comprising 521 genes of up to 10 disjunct time points during activation and different polarization conditions. The gene signatures include central transcriptional regulators of T-cell activation, representing successive waves as well as sustained patterns of induction. They cover sustained repressed, intermediate, and late response expression rates across multiple T-cell populations, thus defining consensus biomarker signatures for T-cell activation. In addition, intermediate and late response activation signatures in CAR T-cell infusion products are correlated to immune effector cell-associated neurotoxicity syndrome.

CONCLUSION

This study is the first to describe temporally resolved gene expression patterns across T-cell populations. These biomarker signatures are a valuable source, e.g., monitoring transcriptional changes during T-cell activation with a reasonable number of genes, annotating T-cell states in single-cell transcriptome studies, or assessing dysregulated functions of human T-cell immunity.

Accurate image derived input function in [18F]SynVesT-1 mouse studies using isoflurane and ketamine/xylazine anesthesia

BACKGROUND

Kinetic modeling in positron emission tomography (PET) requires measurement of the tracer plasma activity in the absence of a suitable reference region. To avoid invasive blood sampling, the use of an image derived input function has been proposed. However, an accurate delineation of the blood pool region in the PET image is necessary to obtain unbiased blood activity. Here, to perform brain kinetic modeling in [18F]SynVesT-1 dynamic scans, we make use of non-negative matrix factorization (NMF) to unmix the activity signal from the different tissues that can contribute to the heart region activity, and extract only the left ventricle activity in an unbiased way. This method was implemented in dynamic [18F]SynVesT-1 scans of mice anesthetized with either isoflurane or ketamine-xylazine, two anesthestics that we showed to affect differently radiotracer kinetics. The left ventricle activity (NMF-IDIF) and a manually delineated cardiac activity (IDIF) were compared with arterial blood samples (ABS), and for isoflurane anesthetized mice, arteriovenous (AV) shunt blood data were compared as well. Finally, brain regional 2 tissue compartment modeling was performed using IDIF and NMF-IDIF, and the model fit accuracy (weighted symmetrical mean absolute percentage error, wsMAPE) as well as the total volume of distribution (VT) were compared.

RESULTS

In isoflurane anesthetized mice, the difference between ABS and NMF-IDIF activity (+ 12.8 [Formula: see text] 11%, p = 0.0023) was smaller than with IDIF (+ 16.4 [Formula: see text] 9.8%, p = 0.0008). For ketamine-xylazine anesthetized mice the reduction in difference was larger (NMF-IDIF: 16.9 [Formula: see text] 10%, p = 0.0057, IDIF: 56.3 [Formula: see text] 14%, p < 0.0001). Correlation coefficient between isoflurane AV-shunt time activity curves and NMF-IDIF (0.97 [Formula: see text] 0.01) was higher than with IDIF (0.94 [Formula: see text] 0.03). The brain regional 2TCM wsMAPE was improved using NMF-IDIF compared with IDIF, in isoflurane (NMF-IDIF: 1.24 [Formula: see text] 0.24%, IDIF: 1.56 [Formula: see text] 0.30%) and ketamine-xylazine (NMF-IDIF: 1.40 [Formula: see text] 0.24, IDIF: 2.62 [Formula: see text] 0.27) anesthetized mice. Finally, brain VT was significantly (p < 0.0001) higher using NMF-IDIF compared with IDIF, in isoflurane (3.97 [Formula: see text] 0.13% higher) and ketamine-xylazine (32.7 [Formula: see text] 2.4% higher) anesthetized mice.

CONCLUSIONS

Image derived left ventricle blood activity calculated with NMF improves absolute activity quantification, and reduces the error in the kinetic modeling fit. These improvements are more pronounced in ketamine-xylazine than in isoflurane anesthetized mice.

Phase recognition in SEM-EDX chemical maps using positive matrix factorization

Images from scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy (EDX) are informative and useful to understand the chemical composition and mixing state of solid materials. Positive matrix factorization (PMF) is a multivariate factor analysis technique that has been used in many applications, and the method is here applied to identify factors that can describe common features between elemental SEM-EDX maps. The procedures of converting both graphics and digital images to PMF input files are introduced, and the PMF analysis is exemplified with an open-access PMF program. A case study of oxygen carrier materials from oxygen carrier aided combustion is presented, and the results show that PMF successfully groups elements into factors, and the maps of these factors are visualized. The produced images provide further information on ash interactions and composition of distinct chemical layers. The method can handle all types of chemical maps and the method is not limited solely to SEM-EDX although these images have been used as an example. The main characteristics of the method are:•Adapting graphics and digital images ready for PMF analysis.•Conversion between 1-D and 2-D datasets allows visualization of common chemical maps of elements grouped in factors.•Handles all types of chemical mappings and large data sets.

Relating mutational signature exposures to clinical data in cancers via signeR 2.0

BACKGROUND

Cancer is a collection of diseases caused by the deregulation of cell processes, which is triggered by somatic mutations. The search for patterns in somatic mutations, known as mutational signatures, is a growing field of study that has already become a useful tool in oncology. Several algorithms have been proposed to perform one or both the following two tasks: (1) de novo estimation of signatures and their exposures, (2) estimation of the exposures of each one of a set of pre-defined signatures.

RESULTS

Our group developed signeR, a Bayesian approach to both of these tasks. Here we present a new version of the software, signeR 2.0, which extends the possibilities of previous analyses to explore the relation of signature exposures to other data of clinical relevance. signeR 2.0 includes a user-friendly interface developed using the R-Shiny framework and improvements in performance. This version allows the analysis of submitted data or public TCGA data, which is embedded in the package for easy access.

CONCLUSION

signeR 2.0 is a valuable tool to generate and explore exposure data, both from de novo or fitting analyses and is an open-source R package available through the Bioconductor project at ( https://doi.org/10.18129/B9.bioc.signeR ).

Automated spatial localization of ankle muscle sites and model-based estimation of joint torque post-stroke via a wearable sensorised leg garment

Assessing a patient's musculoskeletal function during over-ground walking is a primary objective in post-stroke rehabilitation, due to the importance of walking recovery for everyday life. However, the quantitative assessment of musculoskeletal function currently requires lab-constrained equipment, and labor-intensive analyses, which hampers assessment in standard clinical settings. The development of fully wearable systems for the online estimation of muscle-tendon forces and resulting joint torque would aid clinical assessment of motor recovery, it would enhance the detection of neuro-muscular anomalies and it would consequently enable highly personalized treatments. Here, we present a wearable technology that combines (1) a soft garment for the human leg sensorized with 64 flexible and dry electromyography (EMG) electrodes, (2) a generalized and automated algorithm for the localization of leg muscle sites, and (3) an EMG-driven musculoskeletal modeling framework for the estimation of ankle dorsi-plantar flexion torques. Our results showed that the automated clustering algorithm could detect muscle locations in both healthy and post-stroke individuals. The estimated muscle-specific EMG envelopes could be used to drive forward person-specific musculoskeletal models and estimate resulting joint torques accurately across all healthy and post-stroke individuals and across different walking speeds (R2  > 0.82 and RMSD  < 0.16). The technology we proposed opens new avenues for automated muscle localization and quantitative musculoskeletal function assessment during gait in both healthy and neurologically impaired individuals.

Quality indices for topic model selection and evaluation: a literature review and case study

BACKGROUND

Topic models are a class of unsupervised machine learning models, which facilitate summarization, browsing and retrieval from large unstructured document collections. This study reviews several methods for assessing the quality of unsupervised topic models estimated using non-negative matrix factorization. Techniques for topic model validation have been developed across disparate fields. We synthesize this literature, discuss the advantages and disadvantages of different techniques for topic model validation, and illustrate their usefulness for guiding model selection on a large clinical text corpus.

DESIGN, SETTING AND DATA

Using a retrospective cohort design, we curated a text corpus containing 382,666 clinical notes collected between 01/01/2017 through 12/31/2020 from primary care electronic medical records in Toronto Canada.

METHODS

Several topic model quality metrics have been proposed to assess different aspects of model fit. We explored the following metrics: reconstruction error, topic coherence, rank biased overlap, Kendall's weighted tau, partition coefficient, partition entropy and the Xie-Beni statistic. Depending on context, cross-validation and/or bootstrap stability analysis were used to estimate these metrics on our corpus.

RESULTS

Cross-validated reconstruction error favored large topic models (K ≥ 100 topics) on our corpus. Stability analysis using topic coherence and the Xie-Beni statistic also favored large models (K = 100 topics). Rank biased overlap and Kendall's weighted tau favored small models (K = 5 topics). Few model evaluation metrics suggested mid-sized topic models (25 ≤ K ≤ 75) as being optimal. However, human judgement suggested that mid-sized topic models produced expressive low-dimensional summarizations of the corpus.

CONCLUSIONS

Topic model quality indices are transparent quantitative tools for guiding model selection and evaluation. Our empirical illustration demonstrated that different topic model quality indices favor models of different complexity; and may not select models aligning with human judgment. This suggests that different metrics capture different aspects of model goodness of fit. A combination of topic model quality indices, coupled with human validation, may be useful in appraising unsupervised topic models.

Identifying a distinct fibrosis subset of NAFLD via molecular profiling and the involvement of profibrotic macrophages

BACKGROUND

There are emerging studies suggesting that non-alcoholic fatty liver disease (NAFLD) is a heterogeneous disease with multiple etiologies and molecular phenotypes. Fibrosis is the key process in NAFLD progression. In this study, we aimed to explore molecular phenotypes of NAFLD with a particular focus on the fibrosis phenotype and also aimed to explore the changes of macrophage subsets in the fibrosis subset of NAFLD.

METHODS

To assess the transcriptomic alterations of key factors in NAFLD and fibrosis progression, we included 14 different transcriptomic datasets of liver tissues. In addition, two single-cell RNA sequencing (scRNA-seq) datasets were included to construct transcriptomic signatures that could represent specific cells. To explore the molecular subsets of fibrosis in NAFLD based on the transcriptomic features, we used a high-quality RNA-sequencing (RNA-seq) dataset of liver tissues from patients with NAFLD. Non-negative matrix factorization (NMF) was used to analyze the molecular subsets of NAFLD based on the gene set variation analysis (GSVA) enrichment scores of key molecule features in liver tissues.

RESULTS

The key transcriptomic signatures on NAFLD including non-alcoholic steatohepatitis (NASH) signature, fibrosis signature, non-alcoholic fatty liver (NAFL) signature, liver aging signature and TGF-β signature were constructed by liver transcriptome datasets. We analyzed two liver scRNA-seq datasets and constructed cell type-specific transcriptomic signatures based on the genes that were highly expressed in each cell subset. We analyzed the molecular subsets of NAFLD by NMF and categorized four main subsets of NAFLD. Cluster 4 subset is mainly characterized by liver fibrosis. Patients with Cluster 4 subset have more advanced liver fibrosis than patients with other subsets, or may have a high risk of liver fibrosis progression. Furthermore, we identified two key monocyte-macrophage subsets which were both significantly correlated with the progression of liver fibrosis in NAFLD patients.

CONCLUSION

Our study revealed the molecular subtypes of NAFLD by integrating key information from transcriptomic expression profiling and liver microenvironment, and identified a novel and distinct fibrosis subset of NAFLD. The fibrosis subset is significantly correlated with the profibrotic macrophages and M2 macrophage subset. These two liver macrophage subsets may be important players in the progression of liver fibrosis of NAFLD patients.

Evolution of brain network dynamics in early Parkinson's disease with mild cognitive impairment

How mild cognitive impairment (MCI) is instantiated in dynamically interacting and spatially distributed functional brain networks remains an unexplored mystery in early Parkinson's disease (PD). We applied a machine-learning technology based on personalized sliding-window algorithm to track continuously time-varying and overlapping subnetworks under the functional brain networks calculated form resting state electroencephalogram data within a sample of 33 early PD patients (13 early PD patients with MCI and 20 early PD patients without MCI). We decoded a set of subnetworks that captured surprisingly dynamically varying and integrated interactions among certain brain lobes. We observed that the master expressed subnetworks were particularly transient, and flexibly switching between high and low expression during integration into a dynamic brain network. This transience was particularly salient in a subnetwork predominantly linking temporal-parietal-occipital lobes, which decreases in both expression and flexibility in early PD patients with MCI and expresses their degree of cognitive impairment. Moreover, MCI induced a regularly interrupted, slow evolution of subnetworks in functional brain network dynamics in early PD at the individual level, and the dynamic expression characteristics of subnetworks also reflected the degree of cognitive impairment in patients with early PD. Collectively, these results provide novel and deeper insights regarding MCI-induced abnormal dynamical interaction and large-scale changes in functional brain network of early PD.

The sub-molecular characterization identification for cervical cancer

Background

The efficacy of therapy in cervical cancer (CESC) is blocked by high molecular heterogeneity. Thus, the sub-molecular characterization remains primarily explored for personalizing the treatment of CESC patients.

Methods

Datasets with 741 CESC patients were obtained from TCGA and GEO databases. The NMF algorithm, random forest algorithm, and multivariate Cox analysis were utilized to construct a classifier for defining the sub-molecular characterization. Then, the biological characteristics, genomic variations, prognosis, and immune landscape in molecular subtypes were explored. The significance of classifier genes was validated by quantitative Real-Time PCR, cell transfection, cell colony formation assay, wound healing assay, cell proliferation assay, and Western blot.

Results

The CESC patients were classified into two subtypes, and the high classifier-score patients with significant differences in ECM-receptor interaction, PI3K-Akt signaling pathway, and MAPK signaling pathway showed a poorer prognosis in OS (p < 0.001), DFI (p = 0.016), PFI (p < 0.001) and DSS (p < 0.001), and with high the M0 Macrophage and resting Mast cells infiltration and low HLA family gene expression. Moreover, the constructed classifier owns a high identified accuracy in the tumor/normal groups (AUC: 0.993), the tumor/CIN1-CIN3 groups (AUC: 0.963), and normal/CIN1-CIN3 groups (AUC: 0.962), and the total prediction performance is better than currently published signatures in CESC (C-index: 0,763). The combined prediction performance further indicated that Nomogram (AUC = 0.837) is superior to the classifier (AUC = 0.835) and Stage (AUC = 0.568), and the C-index of calibration curves is 0.784. The potential biological function of classifier genes indicated that silencing GALNT2 inhibited the cancer cell's proliferation, migration, and colony formation; Conversely, the cancer cell's proliferation, migration, and colony formation were increased after the upregulation of GALNT2. The Epithelial-Mesenchymal Transition Experiment showed that GALNT2 knockdown might reduce the levels of Snail and Vimentin proteins and increase E-cadherin; Conversely, the levels of Snail and Vimentin proteins were increased, E-cadherin was reduced by GALNT2 upregulation.

Conclusion

The classifier we constructed may help improve our understanding of subtype characteristics and provide a new strategy for developing CESC therapeutics. Remarkably, GALNT2 may be an option to directly target drivers in CESC cancer therapy.

Identifying latent activity behaviors and lifestyles using mobility data to describe urban dynamics

Urbanization and its problems require an in-depth and comprehensive understanding of urban dynamics, especially the complex and diversified lifestyles in modern cities. Digitally acquired data can accurately capture complex human activity, but it lacks the interpretability of demographic data. In this paper, we study a privacy-enhanced dataset of the mobility visitation patterns of 1.2 million people to 1.1 million places in 11 metro areas in the U.S. to detect the latent mobility behaviors and lifestyles in the largest American cities. Despite the considerable complexity of mobility visitations, we found that lifestyles can be automatically decomposed into only 12 latent interpretable activity behaviors on how people combine shopping, eating, working, or using their free time. Rather than describing individuals with a single lifestyle, we find that city dwellers' behavior is a mixture of those behaviors. Those detected latent activity behaviors are equally present across cities and cannot be fully explained by main demographic features. Finally, we find those latent behaviors are associated with dynamics like experienced income segregation, transportation, or healthy behaviors in cities, even after controlling for demographic features. Our results signal the importance of complementing traditional census data with activity behaviors to understand urban dynamics.

Supplementary Information

The online version contains supplementary material available at 10.1140/epjds/s13688-023-00390-w.

Model selection and robust inference of mutational signatures using Negative Binomial non-negative matrix factorization

BACKGROUND

The spectrum of mutations in a collection of cancer genomes can be described by a mixture of a few mutational signatures. The mutational signatures can be found using non-negative matrix factorization (NMF). To extract the mutational signatures we have to assume a distribution for the observed mutational counts and a number of mutational signatures. In most applications, the mutational counts are assumed to be Poisson distributed, and the rank is chosen by comparing the fit of several models with the same underlying distribution and different values for the rank using classical model selection procedures. However, the counts are often overdispersed, and thus the Negative Binomial distribution is more appropriate.

RESULTS

We propose a Negative Binomial NMF with a patient specific dispersion parameter to capture the variation across patients and derive the corresponding update rules for parameter estimation. We also introduce a novel model selection procedure inspired by cross-validation to determine the number of signatures. Using simulations, we study the influence of the distributional assumption on our method together with other classical model selection procedures. We also present a simulation study with a method comparison where we show that state-of-the-art methods are highly overestimating the number of signatures when overdispersion is present. We apply our proposed analysis on a wide range of simulated data and on two real data sets from breast and prostate cancer patients. On the real data we describe a residual analysis to investigate and validate the model choice.

CONCLUSIONS

With our results on simulated and real data we show that our model selection procedure is more robust at determining the correct number of signatures under model misspecification. We also show that our model selection procedure is more accurate than the available methods in the literature for finding the true number of signatures. Lastly, the residual analysis clearly emphasizes the overdispersion in the mutational count data. The code for our model selection procedure and Negative Binomial NMF is available in the R package SigMoS and can be found at https://github.com/MartaPelizzola/SigMoS .

A cross-sectional and longitudinal study of human brain development: The integration of cortical thickness, surface area, gyrification index, and cortical curvature into a unified analytical framework

Brain maturation studies typically examine relationships linking a single morphometric feature with cognition, behavior, age, or other demographic characteristics. However, the coordinated spatiotemporal arrangement of morphological features across development and their associations with behavior are unclear. Here, we examine covariation across multiple cortical features (cortical thickness [CT], surface area [SA], local gyrification index [GI], and mean curvature [MC]) using magnetic resonance images from the NIMH developmental cohort (ages 5-25). Neuroanatomical covariance was examined using non-negative matrix factorization (NMF), which decomposes covariance resulting in a parts-based representation. Cross-sectionally, we identified six components of covariation which demonstrate differential contributions of CT, GI, and SA in hetero- vs. unimodal areas. Using this technique to examine covariance in rates of change to identify longitudinal sources of covariance highlighted preserved SA in unimodal areas and changes in CT and GI in heteromodal areas. Using behavioral partial least squares (PLS), we identified a single latent variable (LV) that recapitulated patterns of reduced CT, GI, and SA related to older age, with limited contributions of IQ and SES. Longitudinally, PLS revealed three LVs that demonstrated a nuanced developmental pattern that highlighted a higher rate of maturational change in SA and CT in higher IQ and SES females. Finally, we situated the components in the changing architecture of cortical gradients. This novel characterization of brain maturation provides an important understanding of the interdependencies between morphological measures, their coordinated development, and their relationship to biological sex, cognitive ability, and the resources of the local environment.

The investigation of an analysis method for co-activation of knee osteoarthritis utilizing normalization of peak dynamic method

BACKGROUND

Assessing co-activation characteristics in knee osteoarthritis (knee OA) using method of quantification of the activity ratio (such as the co-contraction index (CCI) or the directed co-activation ratios (DCAR)) for surface electromyography (EMG) has been reported. However, no studies have discussed the differences in results between non-negative matrix factorization (NNMF) and the DCAR.

RESEARCH QUESTION

Does DCAR or NNMF reflect the characteristic co-activation pattern of knee OA while using EMG normalized by the peak dynamic method?

METHODS

Ten elderly control participants (EC) and ten knee OA patients (KOA) volunteered to participate in this study. EMG data from 20 participants were obtained from our previous study. Patients with knee OA were recruited from a local orthopedic clinic. The DCAR of agonist and antagonist muscles and the number of modules using NNMF were calculated to evaluate multiple muscle co-activations. An independent t-test statistical parametric mapping approach was used to compare the DCAR between the two groups. The difference in the number of modules between EC and KOA was evaluated using the Wilcoxon rank-sum test.

RESULTS

There was no significant difference in the DCAR between the two groups. However, NNMF had significantly fewer modules with KOA than with EC.

SIGNIFICANCE

The NNMF with the ratio of the amplitude of each muscle and duration of activity as variables reflected the co-activation of KOA, characterized by the high synchronous and prolonged activity of each muscle. Therefore, the NNMF is suitable for extracting characteristic muscle activity patterns of knee OA independent of the normalization method.

Continuous estimation of multi-DOF movement from sEMG based on non-negative matrix factorization and L2 regulation

Accurate continuous estimation of multi-DOF movement is crucial for simultaneous control of advanced myoelectric prosthetic. The decoupling of multi-DOF is a challenge for continuous estimation. In this paper, we propose a model combined non-negative matrix factorization (NMF) with Hadamard product and L2 regulation to suppress the non-active DOF and achieve the multi-DOF movement continuous estimation. The L2 regulation of non-active DOF activation coefficient was added to the object function of NMF with the benefit of Hadamard product. The angles were estimated by a linear combination of the activation coefficients. We performed a set of continuous estimation experiments for single-DOF and multi-DOF movements of wrist flexion/extend and hand open/close. The results illustrated that the novel model could suppress non-active DOF in single-DOF movement better than other methods based on muscle synergy theory. Moreover, we investigated the robustness of suppression effect and the similarity of synergy matrices at different speeds for NMF-based methods, and the results showed that the proposed method had a superior performance.

Short text topic modelling using local and global word-context semantic correlation

Nowadays, people use short text to portray their opinions on platforms of social media such as Twitter, Facebook, and YouTube, as well as on e-commerce websites such as Amazon and Flipkart to share their commercial purchasing experiences. Every day, billions of short texts are created worldwide in tweets, tags, keywords, search queries etc. However, this short text possesses inadequate contextual information, which can be ambiguous, sparse, noisy, remains a major challenge. State-of-the-art strategies of topic modeling such as Latent Dirichlet Allocation and Probabilistic Latent Semantic Analysis are not suitable as it contains a limited number of words in a single document. This work proposes a new model named G_SeaNMF (Gensim_SeaNMF) to improve the word-context semantic relationship by using local and global word embedding techniques. Word embeddings learned from a large corpus provide general semantic and syntactic information about words; it can guide topic modeling for short text collections as supporting information for sparse co-occurrence patterns. In the proposed model, SeaNMF (Semantics-assisted Non-negative Matrix Factorization) is incorporated with word2vec model of Gensim library to strengthen the word's semantic relationship. In this article, a short text topic modeling techniques based on DMM (Dirichlet Multinomial Mixture), self-aggregation and global word co-occurrence were explored. These are evaluated using different measures to gauge cluster coherence on real-world datasets such as Search Snippet, Biomedicine, Pascal Flickr, Tweet and TagMyNews. Empirical evaluation shows that a combination of local and global word embedding provides more appropriate words under each topic with improved outcomes.

A weighted non-negative matrix factorization approach to predict potential associations between drug and disease

BACKGROUND

Associations of drugs with diseases provide important information for expediting drug development. Due to the number of known drug-disease associations is still insufficient, and considering that inferring associations between them through traditional in vitro experiments is time-consuming and costly. Therefore, more accurate and reliable computational methods urgent need to be developed to predict potential associations of drugs with diseases.

METHODS

In this study, we present the model called weighted graph regularized collaborative non-negative matrix factorization for drug-disease association prediction (WNMFDDA). More specifically, we first calculated the drug similarity and disease similarity based on the chemical structures of drugs and medical description information of diseases, respectively. Then, to extend the model to work for new drugs and diseases, weighted [Formula: see text] nearest neighbor was used as a preprocessing step to reconstruct the interaction score profiles of drugs with diseases. Finally, a graph regularized non-negative matrix factorization model was used to identify potential associations between drug and disease.

RESULTS

During the cross-validation process, WNMFDDA achieved the AUC values of 0.939 and 0.952 on Fdataset and Cdataset under ten-fold cross validation, respectively, which outperforms other competing prediction methods. Moreover, case studies for several drugs and diseases were carried out to further verify the predictive performance of WNMFDDA. As a result, 13(Doxorubicin), 13(Amiodarone), 12(Obesity) and 12(Asthma) of the top 15 corresponding candidate diseases or drugs were confirmed by existing databases.

CONCLUSIONS

The experimental results adequately demonstrated that WNMFDDA is a very effective method for drug-disease association prediction. We believe that WNMFDDA is helpful for relevant biomedical researchers in follow-up studies.

Interindividual differences in upper limb muscle synergies during baseball throwing motion in male college baseball players

Throwing is a fundamental human motor behavior that has evolved to aid hunting and defense against predators. In modern humans, accurate throwing is an important skill required in many sports. However, the spatiotemporal coordination of muscles during baseball throwing has not been fully elucidated. We herein aimed to identify the muscle synergies involved in baseball throwing and determine whether their spatiotemporal patterns are shared among individuals. Ten college baseball players participated in this study. Electromyographic activity was recorded from 13 ipsilateral upper limb muscles during throwing using full effort. Non-negative matrix factorization was used to extract the motor module composition and temporal activation patterns during baseball throwing, followed by k-means analysis to cluster the extracted motor modules based on their similarity. Four motor modules were extracted for each player. These were classified into four clusters (Clusters 1-4), each reaching the peak activity sequentially from the early cocking phase to ball release. Spatiotemporal interindividual similarity in the muscle synergy cluster comprising the muscles activated during the transition from early cocking to late cocking (Cluster 2) was significantly lower than that in the other clusters. There was no individual-specific muscle synergy. These results suggest that the skilled baseball throwing motion acquired through years of practice may consist of four basic muscle synergies that are common among individuals with some differences in their spatiotemporal patterns.

Automated estimation of ankle muscle EMG envelopes and resulting plantar-dorsi flexion torque from 64 garment-embedded electrodes uniformly distributed around the human leg

The design of personalized movement training and rehabilitation pipelines relies on the ability of assessing the activation of individual muscles concurrently with the resulting joint torques exerted during functional movements. Despite advances in motion capturing, force sensing and bio-electrical recording technologies, the estimation of muscle activation and resulting force still relies on lengthy experimental and computational procedures that are not clinically viable. This work proposes a wearable technology for the rapid, yet quantitative, assessment of musculoskeletal function. It comprises of (1) a soft leg garment sensorized with 64 uniformly distributed electromyography (EMG) electrodes, (2) an algorithm that automatically groups electrodes into seven muscle-specific clusters, and (3) a EMG-driven musculoskeletal model that estimates the resulting force and torque produced about the ankle joint sagittal plane. Our results show the ability of the proposed technology to automatically select a sub-set of muscle-specific electrodes that enabled accurate estimation of muscle excitations and resulting joint torques across a large range of biomechanically diverse movements, underlying different excitation patterns, in a group of eight healthy individuals. This may substantially decrease time needed for localization of muscle sites and electrode placement procedures, thereby facilitating applicability of EMG-driven modelling pipelines in standard clinical protocols.

Road salting and natural brine migration revealed as major sources of groundwater contamination across regions of northern Appalachia with and without unconventional oil and gas development

High methane and salt levels in groundwater have been the most widely cited unconventional oil and gas development (UOGD) related water impairments. The attribution of these contaminants to UOGD is usually complex, especially in regions with mixed land uses. Here, we compiled a large hydrogeochemistry dataset containing 13 geochemical analytes for 17,794 groundwater samples from rural northern Appalachia, i.e., 19 counties located on the boundary between Pennsylvania (PA; UOGD is permitted) and New York (NY; UOGD is banned). With this dataset, we explored if statistical and geospatial tools can help shed light on the sources of inorganic solutes and methane in groundwater in regions with mixed land uses. The traditional Principal Component Analysis (PCA) indicates salts in NY and PA groundwater are mainly from the Appalachian Basin Brine (ABB). In contrast, the machine learning tool - Non-negative Matrix Factorization (NMF) highlights that road salts (in addition to ABB) account for 36%-48% of total chloride in NY and PA groundwaters. The PCA fails to identify road salts as one water/salt source, likely due to its geochemical similarity with ABB. Neither PCA nor NMF detects a regional impact of UOGD on groundwater quality. Our geospatial analyses further corroborate (1) road salting is the major salt source in groundwater, and its impact is enhanced in proximity to highways; (2) UOGD-related groundwater quality deterioration is only limited to a few localities in PA.

Deconvolution of tumor composition using partially available DNA methylation data

Background

Deciphering proportions of constitutional cell types in tumor tissues is a crucial step for the analysis of tumor heterogeneity and the prediction of response to immunotherapy. In the process of measuring cell population proportions, traditional experimental methods have been greatly hampered by the cost and extensive dropout events. At present, the public availability of large amounts of DNA methylation data makes it possible to use computational methods to predict proportions.

Results

In this paper, we proposed PRMeth, a method to deconvolve tumor mixtures using partially available DNA methylation data. By adopting an iteratively optimized non-negative matrix factorization framework, PRMeth took DNA methylation profiles of a portion of the cell types in the tissue mixtures (including blood and solid tumors) as input to estimate the proportions of all cell types as well as the methylation profiles of unknown cell types simultaneously. We compared PRMeth with five different methods through three benchmark datasets and the results show that PRMeth could infer the proportions of all cell types and recover the methylation profiles of unknown cell types effectively. Then, applying PRMeth to four types of tumors from The Cancer Genome Atlas (TCGA) database, we found that the immune cell proportions estimated by PRMeth were largely consistent with previous studies and met biological significance.

Conclusions

Our method can circumvent the difficulty of obtaining complete DNA methylation reference data and obtain satisfactory deconvolution accuracy, which will be conducive to exploring the new directions of cancer immunotherapy. PRMeth is implemented in R and is freely available from GitHub (https://github.com/hedingqin/PRMeth).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-022-04893-7.

uFLIM - Unsupervised analysis of FLIM-FRET microscopy data

Despite their widespread use in cell biology, fluorescence lifetime imaging microscopy (FLIM) data-sets are challenging to analyse, because each spatial position can contain a superposition of multiple fluorescent components. Here, we present a data analysis method employing all information in the available photon budget, as well as being fast. The method, called uFLIM, determines spatial distributions and temporal dynamics of multiple fluorescent components with no prior knowledge. It goes significantly beyond current approaches which either assume the functional dependence of the dynamics, e.g. an exponential decay, or require dynamics to be known, or calibrated. Its efficient non-negative matrix factorization algorithm allows for real-time data processing. We validate in silico that uFLIM is capable to disentangle the spatial distribution and spectral properties of five fluorescing probes, from only two excitation and detection channels and a photon budget of 100 detected photons per pixel. By adapting the method to data exhibiting Förster resonant energy transfer (FRET), we retrieve the spatial and transfer rate distribution of the bound species, without constrains on donor and acceptor dynamics.

Mapping the Heterogeneous Brain Structural Phenotype of Autism Spectrum Disorder Using the Normative Model

BACKGROUND

Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder characterized by substantial clinical and biological heterogeneity. Quantitative and individualized metrics for delineating the heterogeneity of brain structure in ASD are still lacking. Likewise, the extent to which brain structural metrics of ASD deviate from typical development (TD) and whether deviations can be used for parsing brain structural phenotypes of ASD is unclear.

METHODS

T1-weighted magnetic resonance imaging data from the Autism Brain Imaging Data Exchange (ABIDE) II (n_TD = 564) were used to generate a normative model to map brain structure deviations of ABIDE I subjects (n_TD = 560, n_ASD = 496). Voxel-based morphometry was used to compute gray matter volume. Non-negative matrix factorization was employed to decompose the gray matter matrix into 6 factors and weights. These weights were used for normative modeling to estimate the factor deviations. Then, clustering analysis was used to identify ASD subtypes.

RESULTS

Compared with TD, ASD showed increased weights and deviations in 5 factors. Three subtypes with distinct neuroanatomical deviation patterns were identified. ASD subtype 1 and subtype 3 showed positive deviations, whereas ASD subtype 2 showed negative deviations. Distinct clinical manifestations in social communication deficits were identified among the three subtypes.

CONCLUSIONS

Our findings suggest that individuals with ASD have heterogeneous deviation patterns in brain structure. The results highlight the need to test for subtypes in neuroimaging studies of ASD. This study also presents a framework for understanding neuroanatomical heterogeneity in this increasingly prevalent neurodevelopmental disorder.

Sex influence on muscle synergies in a ballistic force-velocity test during the delayed recovery phase after a graded endurance run

The acute and delayed phases of the functional recovery pattern after running exercise have been studied mainly in men. However, it seems that women are less fatigable and/or recover faster than men, at least when tested in isometric condition. After a 20 km graded running race, the influence of sex on the delayed phase of recovery at 2-4 days was studied using a horizontal ballistic force-velocity test. Nine female and height male recreational runners performed maximal concentric push-offs at four load levels a week before the race (PRE), 2 and 4 days (D2 and D4) later. Ground reaction forces and surface electromyographic (EMG) activity from 8 major lower limb muscles were recorded. For each session, the mechanical force-velocity-power profile (i.e. theoretical maximal values of force ( F ¯ 0), velocity ( V ¯ 0), and power ( P ¯ max)) was computed. Mean EMG activity of each recorded muscle and muscle synergies (three for both men and women) were extracted. Independently of the testing sessions, men and women differed regarding the solicitation of the bi-articular thigh muscles (medial hamstring muscles and rectus femoris). At mid-push-off, female made use of more evenly distributed lower limb muscle activities than men. No fatigue effect was found for both sexes when looking at the mean ground reaction forces. However, the force-velocity profile varied by sex throughout the recovery: only men showed a decrease of both V ¯ 0 (p < 0.05) and P ¯ max (p < 0.01) at D2 compared to PRE. Vastus medialis activity was reduced for both men and women up to D4, but only male synergies were impacted at D2: the center of activity of the first and second synergies was reached later. This study suggests that women could recover earlier in a dynamic multi-joint task and that sex-specific organization of muscle synergies may have contributed to their different recovery times after such a race.

Extracting multiple layers of social networks through a 7-month survey using a wearable device: a case study from a farming community in Japan

As individuals are susceptible to social influences from those to whom they are connected, structures of social networks have been an important research subject in social sciences. However, quantifying these structures in real life has been comparatively more difficult. One reason is data collection methods-how to assess elusive social contacts (e.g., unintended brief contacts in a coffee room); however, recent studies have overcome this difficulty using wearable devices. Another reason relates to the multi-layered nature of social relations-individuals are often embedded in multiple networks that are overlapping and complicatedly interwoven. A novel method to disentangle such complexity is needed. Here, we propose a new method to detect multiple latent subnetworks behind interpersonal contacts. We collected data of proximities among residents in a Japanese farming community for 7 months using wearable devices which detect other devices nearby via Bluetooth communication. We performed non-negative matrix factorization (NMF) on the proximity log sequences and extracted five latent subnetworks. One of the subnetworks represented social relations regarding farming activities, and another subnetwork captured the patterns of social contacts taking place in a community hall, which played the role of a "hub" of diverse residents within the community. We also found that the eigenvector centrality score in the farming-related network was positively associated with self-reported pro-community attitude, while the centrality score regarding the community hall was associated with increased self-reported health.

Supplementary Information

The online version contains supplementary material available at 10.1007/s42001-022-00162-y.

Non-Negative Matrix Factorization Temporal Topic Models and Clinical Text Data Identify COVID-19 Pandemic Effects on Primary Healthcare and Community Health in Toronto, Canada

OBJECTIVE

To demonstrate how non-negative matrix factorization can be used to learn a temporal topic model over a large collection of primary care clinical notes, characterizing diverse COVID-19 pandemic effects on the physical/mental/social health of residents of Toronto, Canada.

MATERIALS AND METHODS

The study employs a retrospective open cohort design, consisting of 382,666 primary care progress notes from 44,828 patients, 54 physicians, and 12 clinics collected 01/01/2017 through 31/12/2020. Non-negative matrix factorization uncovers a meaningful latent topical structure permeating the corpus of primary care notes. The learned latent topical basis is transformed into a multivariate time series data structure. Time series methods and plots showcase the evolution/dynamics of learned topics over the study period and allow the identification of COVID-19 pandemic effects. We perform several post-hoc checks of model robustness to increase trust that descriptive/unsupervised inferences are stable over hyper-parameter configurations and/or data perturbations.

RESULTS

Temporal topic modelling uncovers a myriad of pandemic-related effects from the expressive clinical text data. In terms of direct effects on patient-health, topics encoding respiratory disease symptoms display altered dynamics during the pandemic year. Further, the pandemic was associated with a multitude of indirect patient-level effects on topical domains representing mental health, sleep, social and familial dynamics, measurement of vitals/labs, uptake of prevention/screening maneuvers, and referrals to medical specialists. Finally, topic models capture changes in primary care practice patterns resulting from the pandemic, including changes in EMR documentation strategies and the uptake of telemedicine.

CONCLUSION

Temporal topic modelling applied to a large corpus of rich primary care clinical text data, can identify a meaningful topical/thematic summarization which can provide policymakers and public health stakeholders a passive, cost-effective, technology for understanding holistic impacts of the COVID-19 pandemic on the primary healthcare system and community/public-health.

Nondestructive 3D imaging and quantification of hydrated biofilm matrix by confocal Raman microscopy coupled with non-negative matrix factorization

Biofilms are ubiquitous in natural and engineered environments and of great importance in drinking water distribution and biological wastewater treatment systems. Simultaneously acquiring the chemical and structural information of the hydrated biofilm matrix is essential for the cognition and regulation of biofilms in the environmental field. However, the complexity of samples and the limited approaches prevent a holistic understanding of the biofilm matrix. In this work, an approach based on the confocal Raman mapping technique integrated with non-negative matrix factorization (NMF) analysis was developed to probe the hydrated biofilm matrix in situ. The flexibility of the NMF analysis was utilized to subtract the undesired water background signal and resolve the meaningful biological components from Raman spectra of the hydrated biofilms. Diverse chemical components such as proteins, bacterial cells, glycolipids and polyhydroxyalkanoates (PHA) were unraveled within the distinct Pseudomonas spp. biofilm matrices, and the corresponding 3-dimensional spatial organization was visualized and quantified. Of these components, glycolipids and PHA were unique to the P. aeruginosa and P. putida biofilm matrix, respectively. Furthermore, their high abundances in the lower region of the biofilm matrix were found to be related to the specific physiological functions and surrounding microenvironments. Overall, the results demonstrate that our NMF Raman mapping method could serve as a powerful tool complementary to the conventional approaches for identifying and visualizing the chemical components in the biofilm matrix. This work may facilitate the online characterization of the biofilm matrix widely present in the environment and advance the fundamental understanding of biofilm.

Revealing the spatial and temporal distribution of different chemical states of lithium by EELS analysis using non-negative matrix factorization

Detection of the spatial distribution and temporal evolution of an element in different chemical states is difficult in transmission electron microscopy. Here, taking the lithium element as an example, spatial and temporal distribution of different lithium-containing compounds could be revealed by using electron energy-loss spectroscopy (EELS) combined with the analysis method of non-negative matrix factorization (NMF), which is an algorithm that can accomplish the decomposition of high-dimensional data, especially the data which must be positive to implement its physical significance. NMF algorithms of different forms are adopted in this paper to tackle the problem. It is shown that two types of iteration methods, fast hierarchical alternating least squares (Fast-HALS) and spatial orthogonal (SO)-HALS provide decent NMF results on EELS datasets of lithium element. In particular, the low-loss and the core-loss regions of the EELS data are combined together in the process of NMF analysis, enabling better distinction of different chemical states. The above algorithms are recommended for the purpose of analyzing the EELS datasets containing different chemical states of lithium.

Analyses of microstructural variation in the human striatum using non-negative matrix factorization

The striatum is a major subcortical connection hub that has been heavily implicated in a wide array of motor and cognitive functions. Here, we developed a normative multimodal, data-driven microstructural parcellation of the striatum using non-negative matrix factorization (NMF) based on multiple magnetic resonance imaging-based metrics (mean diffusivity, fractional anisotropy, and the ratio between T1- and T2-weighted structural scans) from the Human Connectome Project Young Adult dataset (n = 329 unrelated participants, age range: 22-35, F/M: 185/144). We further explored the biological and functional relationships of this parcellation by relating our findings to motor and cognitive performance in tasks known to involve the striatum as well as demographics. We identified 5 spatially distinct striatal components for each hemisphere. We also show the gain in component stability when using multimodal versus unimodal metrics. Our findings suggest distinct microstructural patterns in the human striatum that are largely symmetric and that relate mostly to age and sex. Our work also highlights the putative functional relevance of these striatal components to different designations based on a Neurosynth meta-analysis.

Assessing Methods for Evaluating the Number of Components in Non-Negative Matrix Factorization

Non-negative matrix factorization is a relatively new method of matrix decomposition which factors an m×n data matrix X into an m×k matrix W and a k×n matrix H, so that X≈W×H. Importantly, all values in X, W, and H are constrained to be non-negative. NMF can be used for dimensionality reduction, since the k columns of W can be considered components into which X has been decomposed. The question arises: how does one choose k? In this paper, we first assess methods for estimating k in the context of NMF in synthetic data. Second, we examine the effect of normalization on this estimate's accuracy in empirical data. In synthetic data with orthogonal underlying components, methods based on PCA and Brunet's Cophenetic Correlation Coefficient achieved the highest accuracy. When evaluated on a well-known real dataset, normalization had an unpredictable effect on the estimate. For any given normalization method, the methods for estimating k gave widely varying results. We conclude that when estimating k, it is best not to apply normalization. If underlying components are known to be orthogonal, then Velicer's MAP or Minka's Laplace-PCA method might be best. However, when orthogonality of the underlying components is unknown, none of the methods seemed preferable.

A comparative study of EOF and NMF analysis on downward trend of AOD over China from 2011 to 2019

In recent decades China has experienced high-level PM_2.5 pollution and then visible air quality improvement. To understand the air quality change from the perspective of aerosol optical depth (AOD), we adopted two statistical methods of Empirical Orthogonal Functions (EOF) and Non-negative Matrix Factorization (NMF) to AOD retrieved by MODIS over China and surrounding areas. Results showed that EOF and NMF identified the important factors influencing AOD over China from different angles: natural dusts controlled the seasonal variation with contribution of 42.4%, and anthropogenic emissions have larger contribution to AOD magnitude. To better observe the interannual variation of different sources, we removed seasonal cycles from original data and conducted EOF analysis on AOD monthly anomalies. Results showed that aerosols from anthropogenic sources had the greatest contribution (27%) to AOD anomaly variation and took an obvious downward trend, and natural dust was the second largest contributor with contribution of 17%. In the areas surrounding China, the eastward aerosol transport due to prevailing westerlies in spring significantly influenced the AOD variation over West Pacific with the largest contribution of 21%, whereas the aerosol transport from BTH region in winter had relative greater impact on the AOD magnitude. After removing seasonal cycles, biomass burning in South Asia became the most important influencing factor on AOD anomalies with contribution of 10%, as its interannual variability was largely affected by El Niño. Aerosol transport from BTH was the second largest contributor with contribution of 8% and showed a decreasing trend. This study showed that the downward trend of AOD over China since 2011 was dominated by aerosols from anthropogenic sources, which in a way confirmed the effectiveness of air pollution control policies.

SC-JNMF: single-cell clustering integrating multiple quantification methods based on joint non-negative matrix factorization

Single-cell RNA-sequencing is a rapidly evolving technology that enables us to understand biological processes at unprecedented resolution. Single-cell expression analysis requires a complex data processing pipeline, and the pipeline is divided into two main parts: The quantification part, which converts the sequence information into gene-cell matrix data; the analysis part, which analyzes the matrix data using statistics and/or machine learning techniques. In the analysis part, unsupervised cell clustering plays an important role in identifying cell types and discovering cell diversity and subpopulations. Identified cell clusters are also used for subsequent analysis, such as finding differentially expressed genes and inferring cell trajectories. However, single-cell clustering using gene expression profiles shows different results depending on the quantification methods. Clustering results are greatly affected by the quantification method used in the upstream process. In other words, even if the original RNA-sequence data is the same, gene expression profiles processed by different quantification methods will produce different clusters. In this article, we propose a robust and highly accurate clustering method based on joint non-negative matrix factorization (joint-NMF) by utilizing the information from multiple gene expression profiles quantified using different methods from the same RNA-sequence data. Our joint-NMF can extract common factors among multiple gene expression profiles by applying each NMF under the constraint that one of the factorized matrices is shared among multiple NMFs. The joint-NMF determines more robust and accurate cell clustering results by leveraging multiple quantification methods compared to conventional clustering methods, which use only a single gene expression profile. Additionally, we showed the usefulness of discovering marker genes with the extracted features using our method.

Primitive muscle synergies reflect different modes of coordination in upper limb motions

The motor system relies on the recruitment of motor modules to perform various movements. Muscle synergies are the modules used by the central nervous system to simplify the control of complex motor tasks. In this paper, we aim to explore the primitive synergies to reflect different modes of coordination in upper limb motions. Muscle synergies and corresponding activation coefficients were extracted via non-negative matrix factorization from the electromyography signals of three basic and four complex upper limb motions in sagittal plane and coronal plane. Similarities of muscle synergies and activation coefficients between different tasks and different subjects were compared. Moreover, we used network theory to assess the coordination between multiple muscles and to elucidate the neural implementation of muscle synergies. The results showed that the combination of different sets of primitive muscle synergies can achieve complex motions in different planes. The muscle synergy network topology differed significantly between different tasks. We also demonstrated the potential of this study for the understanding of human motor control mechanism and implications for neurorehabilitation.

Distinct transcriptional programs stratify ovarian cancer cell lines into the five major histological subtypes

BACKGROUND

Epithelial ovarian cancer (OC) is a heterogenous disease consisting of five major histologically distinct subtypes: high-grade serous (HGSOC), low-grade serous (LGSOC), endometrioid (ENOC), clear cell (CCOC) and mucinous (MOC). Although HGSOC is the most prevalent subtype, representing 70-80% of cases, a 2013 landmark study by Domcke et al. found that the most frequently used OC cell lines are not molecularly representative of this subtype. This raises the question, if not HGSOC, from which subtype do these cell lines derive? Indeed, non-HGSOC subtypes often respond poorly to chemotherapy; therefore, representative models are imperative for developing new targeted therapeutics.

METHODS

Non-negative matrix factorisation (NMF) was applied to transcriptomic data from 44 OC cell lines in the Cancer Cell Line Encyclopedia, assessing the quality of clustering into 2-10 groups. Epithelial OC subtypes were assigned to cell lines optimally clustered into five transcriptionally distinct classes, confirmed by integration with subtype-specific mutations. A transcriptional subtype classifier was then developed by trialling three machine learning algorithms using subtype-specific metagenes defined by NMF. The ability of classifiers to predict subtype was tested using RNA sequencing of a living biobank of patient-derived OC models.

RESULTS

Application of NMF optimally clustered the 44 cell lines into five transcriptionally distinct groups. Close inspection of orthogonal datasets revealed this five-cluster delineation corresponds to the five major OC subtypes. This NMF-based classification validates the Domcke et al. analysis, in identifying lines most representative of HGSOC, and additionally identifies models representing the four other subtypes. However, NMF of the cell lines into two clusters did not align with the dualistic model of OC and suggests this classification is an oversimplification. Subtype designation of patient-derived models by a random forest transcriptional classifier aligned with prior diagnosis in 76% of unambiguous cases. In cases where there was disagreement, this often indicated potential alternative diagnosis, supported by a review of histological, molecular and clinical features.

CONCLUSIONS

This robust classification informs the selection of the most appropriate models for all five histotypes. Following further refinement on larger training cohorts, the transcriptional classification may represent a useful tool to support the classification of new model systems of OC subtypes.

Money flow network among firms' accounts in a regional bank of Japan

In this study, we investigate the flow of money among bank accounts possessed by firms in a region by employing an exhaustive list of all the bank transfers in a regional bank in Japan, to clarify how the network of money flow is related to the economic activities of the firms. The network statistics and structures are examined and shown to be similar to those of a nationwide production network. Specifically, the bowtie analysis indicates what we refer to as a "walnut" structure with core and upstream/downstream components. To quantify the location of an individual account in the network, we used the Hodge decomposition method and found that the Hodge potential of the account has a significant correlation to its position in the bowtie structure as well as to its net flow of incoming and outgoing money and links, namely the net demand/supply of individual accounts. In addition, we used non-negative matrix factorization to identify important factors underlying the entire flow of money; it can be interpreted that these factors are associated with regional economic activities. One factor has a feature whereby the remittance source is localized to the largest city in the region, while the destination is scattered. The other factors correspond to the economic activities specific to different local places. This study serves as a basis for further investigation on the relationship between money flow and economic activities of firms.

Jaya Ant lion optimization-driven Deep recurrent neural network for cancer classification using gene expression data

Cancer is one of the deadly diseases prevailing worldwide and the patients with cancer are rescued only when the cancer is detected at the very early stage. Early detection of cancer is essential as, in the final stage, the chance of survival is limited. The symptoms of cancers are rigorous and therefore, all the symptoms should be studied properly before the diagnosis. Thus, an automatic prediction system is necessary for classifying cancer as malignant or benign. Hence, this paper introduces the novel strategy based on the JayaAnt lion optimization-based Deep recurrent neural network (JayaALO-based DeepRNN) for cancer classification. The steps followed in the developed model are data normalization, data transformation, feature dimension detection, and classification. The first step is data normalization. The goal of data normalization is to eliminate data redundancy and to mitigate the storage of objects in a relational database that maintains the same information in several places. After that, the data transformation is carried out based on log transformation that generates the patterns using more interpretable and helps fulfill the supposition, and to reduce skew. Also, the non-negative matrix factorization is employed for reducing the feature dimension. Finally, the proposed JayaALO-based DeepRNN method effectively classifies cancer based on the reduced dimension features to produce a satisfactory result. Thus, the resulted output of the proposed JayaALO-based DeepRNN is employed for cancer classification. The proposed JayaALO-based DeepRNN showed improved results with maximal accuracy of 95.97%, maximal sensitivity of 95.95%, and maximal specificity of 96.96%. The goal of this research is to devise the cancer classification strategy using the proposed JayaALO-based DeepRNN. It is required to detect the cancer at an early stage to prevent the destruction caused to the other organs. The developed model involves four phases to perform the cancer classification, namely data normalization, data transformation, feature dimension detection, and the classification. Initially, the input images are gathered and are adapted to perform data normalization. The normalized data is fed to the data transformation, which will be performed using log transformation. The obtained transformed data is fed to feature dimension reduction which is performed using non-negative matrix factorization. The reduced features will be employed in DeepRNN for cancer classification. The training of DeepRNN is done using the proposed JayaALO, which is designed by combining ALO and the Jaya algorithm the block diagram of the proposed cancer classification approach using JayaALO-based DeepRNN approach is given below.

Raman imaging reveals in-situ microchemistry of cuticle and epidermis of spruce needles

BACKGROUND

The cuticle is a protective layer playing an important role in plant defense against biotic and abiotic stresses. So far cuticle structure and chemistry was mainly studied by electron microscopy and chemical extraction. Thus, analysing composition involved sample destruction and the link between chemistry and microstructure remained unclear. In the last decade, Raman imaging showed high potential to link plant anatomical structure with microchemistry and to give insights into orientation of molecules. In this study, we use Raman imaging and polarization experiments to study the native cuticle and epidermal layer of needles of Norway spruce, one of the economically most important trees in Europe. The acquired hyperspectral dataset is the basis to image the chemical heterogeneity using univariate (band integration) as well as multivariate data analysis (cluster analysis and non-negative matrix factorization).

RESULTS

Confocal Raman microscopy probes the cuticle together with the underlying epidermis in the native state and tracks aromatics, lipids, carbohydrates and minerals with a spatial resolution of 300 nm. All three data analysis approaches distinguish a waxy, crystalline layer on top, in which aliphatic chains and coumaric acid are aligned perpendicular to the surface. Also in the lipidic amorphous cuticle beneath, strong signals of coumaric acid and flavonoids are detected. Even the unmixing algorithm results in mixed endmember spectra and confirms that lipids co-locate with aromatics. The underlying epidermal cell walls are devoid of lipids but show strong aromatic Raman bands. Especially the upper periclinal thicker cell wall is impregnated with aromatics. At the interface between epidermis and cuticle Calcium oxalate crystals are detected in a layer-like fashion. Non-negative matrix factorization gives the purest component spectra, thus the best match with reference spectra and by this promotes band assignments and interpretation of the visualized chemical heterogeneity.

CONCLUSIONS

Results sharpen our view about the cuticle as the outermost layer of plants and highlight the aromatic impregnation throughout. In the future, developmental studies tracking lipid and aromatic pathways might give new insights into cuticle formation and comparative studies might deepen our understanding why some trees and their needle and leaf surfaces are more resistant to biotic and abiotic stresses than others.

Hierarchical non-negative matrix factorization using clinical information for microbial communities

Background

The human microbiome forms very complex communities that consist of hundreds to thousands of different microorganisms that not only affect the host, but also participate in disease processes. Several state-of-the-art methods have been proposed for learning the structure of microbial communities and to investigate the relationship between microorganisms and host environmental factors. However, these methods were mainly designed to model and analyze single microbial communities that do not interact with or depend on other communities. Such methods therefore cannot comprehend the properties between interdependent systems in communities that affect host behavior and disease processes.

Results

We introduce a novel hierarchical Bayesian framework, called BALSAMICO (BAyesian Latent Semantic Analysis of MIcrobial COmmunities), which uses microbial metagenome data to discover the underlying microbial community structures and the associations between microbiota and their environmental factors. BALSAMICO models mixtures of communities in the framework of nonnegative matrix factorization, taking into account environmental factors. We proposes an efficient procedure for estimating parameters. A simulation then evaluates the accuracy of the estimated parameters. Finally, the method is used to analyze clinical data. In this analysis, we successfully detected bacteria related to colorectal cancer.

Conclusions

These results show that the method not only accurately estimates the parameters needed to analyze the connections between communities of microbiota and their environments, but also allows for the effective detection of these communities in real-world circumstances.

Supplementary Information

The online version contains supplementary material available at (10.1186/s12864-021-07401-y).

Distribution characteristics of methylsiloxanes in atmospheric environment of Saitama, Japan: Diurnal and seasonal variations and emission source apportionment

The large production volume of methylsiloxanes (MSs), combined with their high mobility/volatility and persistence, is a matter of concern from the atmospheric pollution perspective. Therefore, we evaluated of the concentrations and emission sources of MSs, including 7 cyclic methylsiloxanes (D3-D9; CMSs, the number refers to the number of Si-O bonds) and 13 linear methylsiloxanes (L3-L15; LMSs) in ambient air collected from Saitama, Japan. This is a first study regarding the evaluation of 20 methylsiloxanes in the Japanese atmosphere. We improved the air sampling methodology by determination the stability of D5 during a 7-d air sampling and arbitrary sample storage period using polystyrene-divinyl benzene copolymer sorbent (Sep-Pak plus PS-2). We analyzed air samples for MSs seasonally collected from nine locations in Saitama, including urban, suburban, rural, and mountainous areas. The mean CMS and LMS concentrations were 358 ng m^-3 and 13.4 ng m^-3, respectively. The D5 concentrations were distributed widely, with high concentrations in urban/suburban populous areas and dispersed at low concentrations in surrounding areas (north and mountainous areas). We analyzed 7-d air samples collected every week over a year and found apparent seasonal and periodic trends in the CMS concentrations. In the diurnal sampling campaign, we observed periodic fluctuations in ambient CMSs, with an inverse relationship with the atmospheric boundary layer development during the day. Backward trajectories and the prevailing wind direction during the sampling period indicated that the specific profiles of D4 observed in fall/winter weeks and north of Saitama could be ascribed to northwestward air-mass advection. We employed a novel approach in estimating CMSs emission sources and source apportionment by using non-negative matrix factorization (NMF). The concentration matrix was divided successfully into two factors (emission sources) namely, personal care and household products and industrial activities.

Assignment of PM2.5 sources in western Japan by non-negative matrix factorization of concentration-weighted trajectories of GED-ICP-MS/MS element concentrations

Rapid economic growth in Asian countries has raised concerns about the influence of air pollutants transported to Japan by westerly winds. We coupled a gas exchange device (GED) with a tandem inductively coupled plasma mass spectrometer (ICP-MS/MS) to enable direct introduction of PM_2.5 to ICP and thus provide better data than could be obtained from samples collected by conventional filter methods. We used the GED-ICP-MS/MS system in Matsue City in western Japan to monitor in real time 29 elements in PM_2.5 at 10-min intervals and to estimate the pollutant sources by non-negative matrix factorization (NMF) of concentration-weighted air-mass trajectories. The trajectory analysis identified high V, As, Sn, and Sb concentrations over the ocean from Taiwan to Tsushima Strait. NMF analysis revealed that these elements could be decomposed to multiple factors that indicated a large contribution from oceanic areas. The elemental contributions of these factors were high for metals/metalloids with low melting points as oxides, strongly suggesting that they were sourced from combustion of ship fuel. Our results demonstrate that both emissions from ships at sea and land-based emissions from Japan and continental Asia contribute to PM_2.5 in Matsue City.

Exploring the external exposome using wearable passive samplers - The China BAPE study

Environmental exposures are one of the greatest threats to human health, yet we lack tools to answer simple questions about our exposures: what are our personal exposure profiles and how do they change overtime (external exposome), how toxic are these chemicals, and what are the sources of these exposures? To capture variation in personal exposures to airborne chemicals in the gas and particulate phases and identify exposures which pose the greatest health risk, wearable exposure monitors can be deployed. In this study, we deployed passive air sampler wristbands with 84 healthy participants (aged 60-69 years) as part of the Biomarkers for Air Pollutants Exposure (China BAPE) study. Participants wore the wristband samplers for 3 days each month for five consecutive months. Passive samplers were analyzed using a novel gas chromatography high resolution mass spectrometry data-processing workflow to overcome the bottleneck of processing large datasets and improve confidence in the resulting identified features. The toxicity of chemicals observed frequently in personal exposures were predicted to identify exposures of potential concern via inhalation route or other routes of airborne contaminant exposure. Three exposures were highlighted based on elevated toxicity: dichlorvos from insecticides (mosquito/malaria control), naphthalene partly from mothballs, and 183 polyaromatic hydrocarbons from multiple sources. Other exposures explored in this study are linked to diet and personal care products, cigarette smoke, sunscreen, and antimicrobial soaps. We highlight the potential for this workflow employing wearable passive samplers for prioritizing chemicals of concern at both the community and individual level, and characterizing sources of exposures for follow up interventions.

Sparse robust graph-regularized non-negative matrix factorization based on correntropy

Non-negative Matrix Factorization (NMF) is a popular data dimension reduction method in recent years. The traditional NMF method has high sensitivity to data noise. In the paper, we propose a model called Sparse Robust Graph-regularized Non-negative Matrix Factorization based on Correntropy (SGNMFC). The maximized correntropy replaces the traditional minimized Euclidean distance to improve the robustness of the algorithm. Through the kernel function, correntropy can give less weight to outliers and noise in data but give greater weight to meaningful data. Meanwhile, the geometry structure of the high-dimensional data is completely preserved in the low-dimensional manifold through the graph regularization. Feature selection and sample clustering are commonly used methods for analyzing genes. Sparse constraints are applied to the loss function to reduce matrix complexity and analysis difficulty. Comparing the other five similar methods, the effectiveness of the SGNMFC model is proved by selection of differentially expressed genes and sample clustering experiments in three The Cancer Genome Atlas (TCGA) datasets.

Time-resolved compositional mapping during in situ TEM studies

In situ studies using transmission electron microscopy (TEM) can provide insights to how properties, structures and compositions of nanostructures are affected and evolving when exerted to heat or chemical exposure. While high-resolved imaging can be obtained continuously, at video-framerates of hundreds of frames per second (fps), compositional analysis struggles with time resolution due to the long acquisition times for a reliable analysis. This especially holds true when performing mapping (correlated spatial and compositional information). Hence, transient changes are difficult to resolve using mapping. In this work, the time-resolution of sequential mapping using scanning TEM (STEM) and energy dispersive spectroscopy (EDS) is improved by acquiring spectrum images during short times and filtering the spectroscopic data. The suggested algorithm uses regularization to smooth and prevent overfitting (known from compressed sensing) to fit model spectra to the data. The algorithm is applied on simulations as well as acquisitions of catalyzed crystal growth (nanowires), performed in situ in a specialized environmental TEM (ETEM). The results show the improved temporal resolution, where the compositional progression of the different regions of the nanostructure is revealed, here with a time-resolution as low as 16 s compared to the minutes usually needed for similar analysis.

Non-negative matrix factorization for mining big data obtained using four-dimensional scanning transmission electron microscopy

Scientific instruments for material characterization have recently been improved to yield big data. For instance, scanning transmission electron microscopy (STEM) allows us to acquire many diffraction patterns from a scanning area, which is referred to as four-dimensional (4D) STEM. Here we study a combination of 4D-STEM and a statistical technique called non-negative matrix factorization (NMF) to deduce sparse diffraction patterns from a 4D-STEM data consisting of 10,000 diffraction patterns. Titanium oxide nanosheets are analyzed using this combined technique, and we discriminate the two diffraction patterns from pristine TiO₂ and reduced Ti₂O₃ areas, where the latter is due to topotactic reduction induced by electron irradiation. The combination of NMF and 4D-STEM is expected to become a standard characterization technique for a wide range materials.