Standard Clustering Research Articles

Clustering and Arnoux-Rauzy words

We characterize the clustering of a word under the Burrows-Wheeler transform in terms of the resolution of a bounded number of bispecial factors belonging to the language generated by all its powers. We use this criterion to compute, in every given Arnoux-Rauzy language on three letters, an explicit bound K such that each word of length at least K is not clustering; this bound is sharp for a set of Arnoux-Rauzy languages including the Tribonacci one. In the other direction, we characterize all standard Arnoux-Rauzy clustering words, and all perfectly clustering Arnoux-Rauzy words. We extend some results to episturmian languages, characterizing those which produce infinitely many clustering words, and to larger alphabets.

A model-based clustering algorithm with covariates adjustment and its application to lung cancer stratification.

Usually, the clustering process is the first step in several data analyses. Clustering allows identify patterns we did not note before and helps raise new hypotheses. However, one challenge when analyzing empirical data is the presence of covariates, which may mask the obtained clustering structure. For example, suppose we are interested in clustering a set of individuals into controls and cancer patients. A clustering algorithm could group subjects into young and elderly in this case. It may happen because the age at diagnosis is associated with cancer. Thus, we developed CEM-Co, a model-based clustering algorithm that removes/minimizes undesirable covariates' effects during the clustering process. We applied CEM-Co on a gene expression dataset composed of 129 stage I non-small cell lung cancer patients. As a result, we identified a subgroup with a poorer prognosis, while standard clustering algorithms failed.

Combined proteomics and transcriptomics identifies serpin family C member 1 associated protein as a biomarker of endometriosis

Objective To explore potential biomarkers indicating endometriosis (EM). Materials and methods A proteomics method and combined quantitative transcriptomics were adopted to highlight markers in the EM. Venn analysis was used to integrate the ribonucleic acid sequencing (RNA-seq) and protein profiles. Promising candidate markers were tested by enzyme-related immunosorbent assay. Results A sum of 979 mRNAs and 39 proteins were tested to be significantly differentially expression in the standard cluster compared with the EM cluster. Venn analysis showed a filtered signature of only two down-regulated molecules in the EM group, i.e. fetuin B (FETUB) and serpin family C member 1 (SERPINC1); the latter showed a big variance between the control category and the EM set in the authentication test. Conclusion SERPINC1 may be a useful possible biomarker for the analysis of EM.

The value of single-channel endoscopic traction and kiss suture technique in closing wounds caused by endoscopic resection of gastrointestinal muscularis propria tumors.

To investigate the value of single forceps endoscopic traction stapling suture technique (SFETSST) in closing wounds caused by endoscopic resection of gastrointestinal muscularis propria tumor (GMPT). Consecutive patients who underwent submucosal tumor excavation (ESE) and endoscopic full-thickness resection (EFR) for GMPT in the Second Affiliated Hospital of Xiamen Medical College from January 2015 to January 2022 were retrospectively collected. They were divided into the SFETSST group and the standard group (patients who receive single forceps traction-free endoscopic suture technique). The healing effects were compared between the two groups. Seventy-seven patients were included in our study with 50 patients included in SFETSST group. The baseline characteristics had no significant difference between the two groups. The technical success rate of wound suture in SFETSST cluster was significantly upper than that within standard cluster (100% vs. 88.89%, P = 0.04). The wound suture time in SFETSST cluster was significantly lower than that within standard cluster (33.19 ± 10.64min, P < 0.001). Moreover, the incidence rates of intra-operative and postoperative complications in SFETSST cluster were lower than standard cluster (0 vs. 7.41%, P = 0.051 and 0 vs. 11.11%, P = 0.016). Interestingly, the SFETSST cluster had lower cost of consumables (2485.40 ± 591.78 vs. 4098.52 ± 1903.06 Yuan, P = 0.01) and shorter hospital stay (4.96 ± 0.90 vs. 7.19 ± 2.45, P < 0.001) than standard cluster. Our study showed that to fully closure the full-thickness defects of digestive tract, SFETSST was effective, safe, and economical, which was worth popularizing.

Identification of cell subpopulations associated with disease phenotypes from scRNA-seq data using PACSI

BackgroundSingle-cell RNA sequencing (scRNA-seq) has revolutionized the transcriptomics field by advancing analyses from tissue-level to cell-level resolution. Despite the great advances in the development of computational methods for various steps of scRNA-seq analyses, one major bottleneck of the existing technologies remains in identifying the molecular relationship between disease phenotype and cell subpopulations, where “disease phenotype” refers to the clinical characteristics of each patient sample, and subpopulation refer to groups of single cells, which often do not correspond to clusters identified by standard single-cell clustering analysis. Here, we present PACSI, a method aimed at distinguishing cell subpopulations associated with disease phenotypes at the single-cell level.ResultsPACSI takes advantage of the topological properties of biological networks to introduce a proximity-based measure that quantifies the correlation between each cell and the disease phenotype of interest. Applied to simulated data and four case studies, PACSI accurately identified cells associated with disease phenotypes such as diagnosis, prognosis, and response to immunotherapy. In addition, we demonstrated that PACSI can also be applied to spatial transcriptomics data and successfully label spots that are associated with poor survival of breast carcinoma.ConclusionsPACSI is an efficient method to identify cell subpopulations associated with disease phenotypes. Our research shows that it has a broad range of applications in revealing mechanistic and clinical insights of diseases.

Changes in trachoma indicators in Kiribati with two rounds of azithromycin mass drug administration, measured in serial population-based surveys.

Baseline mapping in the two major population centers of Kiribati showed that trachoma was a public health problem in need of programmatic interventions. After conducting two annual rounds of antibiotic mass drug administration (MDA), Kiribati undertook trachoma impact surveys in 2019, using standardized two-stage cluster surveys in the evaluation units of Kiritimati Island and Tarawa. In Kiritimati, 516 households were visited and in Tarawa, 772 households were visited. Nearly all households had a drinking water source and access to an improved latrine. The prevalence of trachomatous trichiasis remained above the elimination threshold (0.2% in ≥15-year-olds) and was virtually unchanged from baseline. The prevalence of trachomatous inflammation-follicular (TF) in 1-9-year-olds decreased by approximately 40% from baseline in both evaluation units but remained above the 5% TF prevalence threshold for stopping MDA. TF prevalence at impact survey was 11.5% in Kiritimati and 17.9% in Tarawa. Infection prevalence in 1-9-year-olds by PCR was 0.96% in Kiritimati and 3.3% in Tarawa. Using a multiplex bead assay to measure antibodies to the C. trachomatis antigen Pgp3, seroprevalence in 1-9-year-olds was 30.2% in Kiritimati and 31.4% in Tarawa. The seroconversion rate, in seroconversion events/100 children/year, was 9.0 in Kiritimati and 9.2 in Tarawa. Seroprevalence and seroconversion rates were both assessed by four different assays, with strong agreement between tests. These results show that, despite decreases in indicators associated with infection at impact survey, trachoma remains a public health problem in Kiribati, and provide additional information about changes in serological indicators after MDA.

Effect of graphene on solid–liquid coexistence in Cu nanodroplets

The microstructure determines the macro properties, so the chemical and physical properties of nanoparticles can be tuned by changing their structures. A molecular dynamics simulation for the rapid solidification of Cu nanodroplets on the graphene sheet is conducted to investigate the effect of substrate on nanoparticles. The phase transition, structure evolution, and crystallinity are quantified by the potential energy, the largest standard cluster analysis, and Bond-orientation order parameters. It is found that the solidification is inhomogeneous and stepwise crystallization. Interestingly, originating from the graphene wettability, different solid–liquid coexistence states are obtained that may have the potential to modify the chemical and physical reactions on nanoparticles. The temperature of this solid–liquid coexistence state depends strongly on the structural evolution of crystallization and thus can be regulated by the wetting properties of the graphene substrate. These results significantly extend the fundamental understanding of heterogeneous nucleation in nanodroplets, providing important insights for designing and fabricating composite materials with multifunctional properties.

Fuzzy-Rough induced spectral ensemble clustering

Ensemble clustering helps achieve fast clustering under abundant computing resources by constructing multiple base clusterings. Compared with the standard single clustering algorithm, ensemble clustering integrates the advantages of multiple clustering algorithms and has stronger robustness and applicability. Nevertheless, most ensemble clustering algorithms treat each base clustering result equally and ignore the difference of clusters. If a cluster in a base clustering is reliable/unreliable, it should play a critical/uncritical role in the ensemble process. Fuzzy-rough sets offer a high degree of flexibility in enabling the vagueness and imprecision present in real-valued data. In this paper, a novel fuzzy-rough induced spectral ensemble approach is proposed to improve the performance of clustering. Specifically, the significance of clusters is differentiated, and the unacceptable degree and reliability of clusters formed in base clustering are induced based on fuzzy-rough lower approximation. Based on defined cluster reliability, a new co-association matrix is generated to enhance the effect of diverse base clusterings. Finally, a novel consensus spectral function is defined by the constructed adjacency matrix, which can lead to significantly better results. Experimental results confirm that the proposed approach works effectively and outperforms many state-of-the-art ensemble clustering algorithms and base clustering, which illustrates the superiority of the novel algorithm.

Synonyms, Antonyms and Factual Knowledge in BERT Heads

In recent years, many studies have been devoted to discovering the inner workings of Transformer-based models, such as BERT, for instance, attempting to identify what information is contained within them. However, little is known about how these models store this information in their millions of parameters and which parts of the architecture are the most important. In this work, we propose an approach to identify self-attention mechanisms, called heads, that contain semantic and real-world factual knowledge in BERT. Our approach includes a metric computed from attention weights and exploits a standard clustering algorithm for extracting the most relevant connections between tokens in a head. In our experimental analysis, we focus on how heads can connect synonyms, antonyms and several types of factual knowledge regarding subjects such as geography and medicine.

Enhanced functional connectome of cerebellum in chronic insomnia patients.

Functional abnormalities of the cerebellum have been found to be closely associated with chronic insomnia (CI). However, whether there are abnormalities in the topology of the functional connectome of the cerebellum in these patients is still unknown. This study aimed to investigate topological abnormalities of the cerebellar functional connectome in individuals with CI. We used resting-state functional magnetic resonance imaging (rs-fMRI) and graph-theoretic analysis to construct a functional connectivity matrix and extract topological property features of the cerebellar functional connectome in patients with CI. We examined global and nodal topological property changes in the cerebellar functional connectome in 102 patients with CI (CI group) and 101 healthy participants without insomnia symptoms (HC group) to determine the differences between groups. Correlations between the topological properties of the cerebellar functional connectome and clinical assessments were calculated to confirm the differences between groups. The cerebellar functional connectome of both CI and HC patients exhibited small-world properties. The CI group showed higher standardized clustering coefficients at the global properties and higher betweenness centrality in the cerebellar Crus II vermis region at the nodal properties compared with participants in the HC group. However, the topological properties of cerebellar functional connectome abnormalities in the CI group were not significantly different from those in clinical assessments. Our findings suggest that the abnormal global and nodal topological properties of the cerebellar functional connectome are associated with CI and could serve as an important biomarker for CI.

CIARA: a cluster-independent algorithm for identifying markers of rare cell types from single-cell sequencing data.

A powerful feature of single-cell genomics is the possibility of identifying cell types from their molecular profiles. In particular, identifying novel rare cell types and their marker genes is a key potential of single-cell RNA sequencing. Standard clustering approaches perform well in identifying relatively abundant cell types, but tend to miss rarer cell types. Here, we have developed CIARA (Cluster Independent Algorithm for the identification of markers of RAre cell types), a cluster-independent computational tool designed to select genes that are likely to be markers of rare cell types. Genes selected by CIARA are subsequently integrated with common clustering algorithms to single out groups of rare cell types. CIARA outperforms existing methods for rare cell type detection, and we use it to find previously uncharacterized rare populations of cells in a human gastrula and among mouse embryonic stem cells treated with retinoic acid. Moreover, CIARA can be applied more generally to any type of single-cell omic data, thus allowing the identification of rare cells across multiple data modalities. We provide implementations of CIARA in user-friendly packages available in R and Python.

Cluster-based ensemble learning model for improving sentiment classification of Arabic documents

AbstractThis article reports on designing and implementing a multiclass sentiment classification approach to handle the imbalanced class distribution of Arabic documents. The proposed approach, sentiment classification of Arabic documents (SCArD), combines the advantages of a clustering-based undersampling (CBUS) method and an ensemble learning model to aid machine learning (ML) classifiers in building accurate models against highly imbalanced datasets. The CBUS method applies two standard clustering algorithms:K-means and expectation–maximization, to balance the ratio between the major and the minor classes by decreasing the number of the major class instances and maintaining the number of the minor class instances at the cluster level. The merits of the proposed approach are that it does not remove the majority class instances from the dataset nor injects the dataset with artificial minority class instances. The resulting balanced datasets are used to train two ML classifiers, random forest and updateable Naïve Bayes, to develop prediction data models. The best prediction data models are selected based on F1-score rates. We applied two techniques to test SCArD and generate new predictions from the imbalanced test dataset. The first technique uses the best prediction data models. The second technique uses the majority voting ensemble learning model, which combines the best prediction data models to generate the final predictions. The experimental results showed that SCArD is promising and outperformed the other comparative classification models based on the F1-score rates.

Study protocol for The GOAL Trial: comprehensive geriatric assessment for frail older people with chronic kidney disease to increase attainment of patient-identified goals—a cluster randomised controlled trial

BackgroundAn increasing number of older people are living with chronic kidney disease (CKD). Many have complex healthcare needs and are at risk of deteriorating health and functional status, which can adversely affect their quality of life. Comprehensive geriatric assessment (CGA) is an effective intervention to improve survival and independence of older people, but its clinical utility and cost-effectiveness in frail older people living with CKD is unknown.MethodsThe GOAL Trial is a pragmatic, multi-centre, open-label, superiority, cluster randomised controlled trial developed by consumers, clinicians, and researchers. It has a two-arm design, CGA compared with standard care, with 1:1 allocation of a total of 16 clusters. Within each cluster, study participants ≥ 65 years of age (or ≥ 55 years if Aboriginal or Torres Strait Islander (First Nations Australians)) with CKD stage 3–5/5D who are frail, measured by a Frailty Index (FI) of > 0.25, are recruited. Participants in intervention clusters receive a CGA by a geriatrician to identify medical, social, and functional needs, optimise medication prescribing, and arrange multidisciplinary referral if required. Those in standard care clusters receive usual care. The primary outcome is attainment of self-identified goals assessed by standardised Goal Attainment Scaling (GAS) at 3 months. Secondary outcomes include GAS at 6 and 12 months, quality of life (EQ-5D-5L), frailty (Frailty Index – Short Form), transfer to residential aged care facilities, cost-effectiveness, and safety (cause-specific hospitalisations, mortality). A process evaluation will be conducted in parallel with the trial including whether the intervention was delivered as intended, any issue or local barriers to intervention delivery, and perceptions of the intervention by participants. The trial has 90% power to detect a clinically meaningful mean difference in GAS of 10 units.DiscussionThis trial addresses patient-prioritised outcomes. It will be conducted, disseminated and implemented by clinicians and researchers in partnership with consumers. If CGA is found to have clinical and cost-effectiveness for frail older people with CKD, the intervention framework could be embedded into routine clinical practice. The implementation of the trial’s findings will be supported by presentations at conferences and forums with clinicians and consumers at specifically convened workshops, to enable rapid adoption into practice and policy for both nephrology and geriatric disciplines. It has potential to materially advance patient-centred care and improve clinical and patient-reported outcomes (including quality of life) for frail older people living with CKD.Trial registrationClinicalTrials.gov NCT04538157. Registered on 3 September 2020.

Optimization of Crop Recommendations Using Novel Machine Learning Techniques

A farmer can use machine learning to make decisions about what crops to sow, how to care for those crops throughout the growing season, and how to predict crop yields. According to the World Health Organization, agriculture is essential to the nation’s quick economic development. Food security, access, and adoption are the three cornerstones of the organization. Without a doubt, the main priority is to ensure that there is enough food for everyone. Increasing agricultural yield can help ensure a sufficient supply. The country-wide variation in crop yields is substantial. As a result, this will be the foundation for research into whether cluster analysis can be used to identify crop yield patterns in a field. Previous study investigations were only marginally successful in accomplishing their primary intended objectives because of unstable conditions and imprecise methodology. The vast majority of farmers base their predictions of crop yield on prior observations of crop growth in their farms, which can be deceptive. Standard preprocessing methods and random cluster value selection are not always reliable, according to the literature. The proposed study overcomes the shortcomings of conventional methodology by highlighting the significance of machine learning-based classification/partitioning and hierarchical approaches in offering a trained analysis of yield prediction in the state of Karnataka. The dataset used for the study was collected from the ICAR-Taralabalu Krishi Vigyan Kendra, Davangere, Karnataka. In the two dataset analysis techniques employed in the study to find anomalies, crop area, and crop production are significant variables. Crop area and crop yield are important variables in the two dataset analysis methods used in the study to detect anomalies. The study emphasizes the importance of a mathematical model and algorithm for identifying yield trends, which can assist farmers in selecting crops that have a large seasonal impact on yield productivity.

CIM-powered physics-based assessment of wind damages to building clusters considering trees

It is important to obtain accurate predictions of wind damage to building clusters for reducing wind-induced economic losses. To this end, a city information model-powered physics-based method for assessing the wind damage to building clusters considering trees was proposed. The influence of trees on wind damage was investigated based on two virtual standard building clusters, revealing that trees can play a shielding role to reduce the risk level of buildings. The reliability and feasibility of the method were demonstrated based on a case study of a manufacturing factory. Moreover, the applicability of the method in city-scale applications was illustrated based on a large urban area covering 619 buildings and 19,740 trees. Based on meteorological forecasting data, the proposed method can help city managers identify and reinforce the building with a high damage risk before strong wind events, thus promoting a safer city.

Capturing the dynamics of microbial interactions through individual-specific networks.

Longitudinal analysis of multivariate individual-specific microbiome profiles over time or across conditions remains dauntin. Most statistical tools and methods that are available to study microbiomes are based on cross-sectional data. Over the past few years, several attempts have been made to model the dynamics of bacterial species over time or across conditions. However, the field needs novel views on handling microbial interactions in temporal analyses. This study proposes a novel data analysis framework, MNDA, that combines representation learning and individual-specific microbial co-occurrence networks to uncover taxon neighborhood dynamics. As a use case, we consider a cohort of newborns with microbiomes available at 6 and 9 months after birth, and extraneous data available on the mode of delivery and diet changes between the considered time points. Our results show that prediction models for these extraneous outcomes based on an MNDA measure of local neighborhood dynamics for each taxon outperform traditional prediction models solely based on individual-specific microbial abundances. Furthermore, our results show that unsupervised similarity analysis of newborns in the study, again using the notion of a taxon's dynamic neighborhood derived from time-matched individual-specific microbial networks, can reveal different subpopulations of individuals, compared to standard microbiome-based clustering, with potential relevance to clinical practice. This study highlights the complementarity of microbial interactions and abundances in downstream analyses and opens new avenues to personalized prediction or stratified medicine with temporal microbiome data.

Volcanic earthquake catalog enhancement using integrated detection, matched-filtering, and relocation tools

Volcanic earthquake catalogs are an essential data product used to interpret subsurface volcanic activity and forecast eruptions. Advances in detection techniques (e.g., matched-filtering, machine learning) and relative relocation tools have improved catalog completeness and refined event locations. However, most volcano observatories have yet to incorporate these techniques into their catalog-building workflows. This is due in part to complexities in operationalizing, automating, and calibrating these techniques in a satisfactory way for disparate volcano networks and their varied seismicity. In an effort to streamline the integration of catalog-enhancing tools at the Alaska Volcano Observatory (AVO), we have integrated four popular open-source tools: REDPy, EQcorrscan, HypoDD, and GrowClust. The combination of these tools offers the capability of adding seismic event detections and relocating events in a single workflow. The workflow relies on a combination of standard triggering and cross-correlation clustering (REDPy) to consolidate representative templates used in matched-filtering (EQcorrscan). The templates and their detections are then relocated using the differential time methods provided by HypoDD and/or GrowClust. Our workflow also provides codes to incorporate campaign data at appropriate junctures, and calculate magnitude and frequency index for valid events. We apply this workflow to three datasets: the 2012–2013 seismic swarm sequence at Mammoth Mountain (California), the 2009 eruption of Redoubt Volcano (Alaska), and the 2006 eruption of Augustine Volcano (Alaska); and compare our results with previous studies at each volcano. In general, our workflow provides a significant increase in the number of events and improved locations, and we relate the event clusters and temporal progressions to relevant volcanic activity. We also discuss workflow implementation best practices, particularly in applying these tools to sparse volcano seismic networks. We envision that our workflow and the datasets presented here will be useful for detailed volcano analyses in monitoring and research efforts.

A New Multistage Clustering Algorithm for Optimal VAr Planning For Dynamic Voltage Stability Analysis

Power systems need reactive power support to withstand voltage instability issues. The Volt Ampere reactive (VAr) resources are to be optimally determined for location and size. Generally, candidate locations are determined based on the ranking of buses according to sensitivity indices, and the Trajectory Sensitivity Index (TSI) is the most widely used sensitivity index. However, the locations identified by ranking the buses at the system level using the TSI method do not guarantee an optimal VAr solution. The recent methods based on dividing the power system into zones address the issue of optimality. The prevailing zoning methods are based on the computation of the electrical distance between buses and do not address the zoning scenario where the electrical distance between the buses is indeterminable. This paper introduces a new multistage algorithm for grouping the buses into zones even when the electrical distance is indeterminable. The new proposed strategy encompasses applying proposed analytical and parametric techniques after the standard clustering steps. A new index, namely the Bilateral Sensitivity Index (BSI), is introduced to quantify the proximity of buses. The TSI-based zonal level ranking of buses is introduced. The proposed strategy demonstrates that when TSI is employed to rank the buses at the zonal level rather than at the system level, it helps towards an optimal VAr solution. The new multistage clustering strategy is integrated into the generalized Dynamic Voltage Stability Analysis (DVSA) framework for optimal VAr determination and applied to study the voltage collapse phenomenon on a practical 24-bus system in the southern part of India. The results demonstrate the proposed approach’s efficacy in mitigating voltage instability issues through optimal VAr support.

Manual cell selection in single cell transcriptomics using scSELpy supports the analysis of immune cell subsets.

Single cell RNA sequencing plays an increasing and indispensable role in immunological research such as in the field of inflammatory bowel diseases (IBD). Professional pipelines are complex, but tools for the manual selection and further downstream analysis of single cell populations are missing so far. We developed a tool called scSELpy, which can easily be integrated into Scanpy-based pipelines, allowing the manual selection of cells on single cell transcriptomic datasets by drawing polygons on various data representations. The tool further supports the downstream analysis of the selected cells and the plotting of results. Taking advantage of two previously published single cell RNA sequencing datasets we show that this tool is useful for the positive and negative selection of T cell subsets implicated in IBD beyond standard clustering. We further demonstrate the feasibility for subphenotyping T cell subsets and use scSELpy to corroborate earlier conclusions drawn from the dataset. Moreover, we also show its usefulness in the context of T cell receptor sequencing. Collectively, scSELpy is a promising additive tool fulfilling a so far unmet need in the field of single cell transcriptomic analysis that might support future immunological research.

An improved differential bond energy algorithm with fuzzy merging method to improve the document clustering for information mining

AbstractThe vast and diversified text materials on the internet in recent years have drastically increased the importance of information mining. By organizing documents into cohesive groupings, a document clustering method is a suitable tool for dealing with massive amounts of documents. Text documents, on the other hand, include sparse and uninformative features like noise, unrelated, and unneeded features, which reduce the efficiency of the document clustering methods. For noise removal, this work employs the hyperclique pattern‐based data cleaner pre‐processing approach. Then, differential bond energy algorithm (DBEA) is combined with a fuzzy merging approach and termed improved differential bond energy algorithm with fuzzy merging (IDBEFM) to handle the issues present in text document clustering. It seeks to discover and display natural variable clusters among large amounts of data. IDBEFM clusters are documented in three steps: in the first step, a cluster similarity matrix is instantiated by applying the DBEA. The second step develops an innovative technique for automatically dividing the cluster matrix into compact cohesive clusters, and in the third step, a fuzzy merging approach is used for combining identical clusters by the correlations and interrelationships between the resultant clusters. The test findings demonstrated that the accomplishment of the proposed technique is much superior to standard clustering algorithms.