Community detection algorithms: a comparative analysis

Research on Multi-Scale Ecological Network Connectivity-Taking the Guangdong-Hong Kong-Macao Greater Bay Area as a Case Study

The Guangdong-Hong Kong-Macao Greater Bay Area urban agglomeration is an urban agglomeration with some of the most intensive urbanization since 1980s. A large amount of cultivated land, forest land, water bodies and other land types in the region has been occupied by construction land, resulting in fragmented ecological landscapes and biodiversity in the region and causing many other ecological problems. Based on this, this paper takes the Guangdong-Hong Kong-Macao Greater Bay Area as a case study, constructs an ecological network of the dispersion scale of five species from 1990 to 2020 based on a morphological spatial pattern analysis (MSPA) method, identifies the ecological groups in the network and uses the core node-based community evolution path tracking algorithm to analyze the ecological groups in order to explore the changes of ecological network connectivity at different scales in the region and to reveal the overall and local characteristics and changes of the migratory space of terrestrial mammals with different dispersion capabilities. The research results show that: (1) From 1990 to 2020, the area of construction land in the Guangdong-Hong Kong-Macao Greater Bay Area increased sharply, with good connectivity in the northwest, southwest and eastern regions and poor connectivity in the central region. (2) There are obvious differences between the overall and local changes in the connectivity trends of multi-scale regional ecological networks. On the whole, the overall ecological connectivity of the ecological network at each scale showed a gradual upward trend, and the overall connectivity index IIC and the possible connectivity index PC gradually increased with the increase of the maximum dispersal distance of species. From the perspective of local patches, the larger the species dispersion scale, the larger the value of the revised betweenness centrality index and the patch possible connectivity index. (3) The distribution of ecological groups at different species dispersion scales is different, and the smaller the dispersal scale of the species, the greater the distribution of ecological groups. Small-scale species are limited by the maximum dispersal distance, and the range of their ecological groups is generally small. Small-scale (3 km), mesoscale (10 km) and large-scale (30 km) core nodes of ecological groups show a gradual increase trend, and the overall connectivity of ecological groups has improved. However, the core nodes of the extra-large-scale (60 km) and ultra-large-scale (100 km) ecological groups show a trend of decreasing fluctuations, and the overall connectivity within the ecological group has declined. This study is helpful to clarify the structural characteristics of regional ecological space and provide a theoretical basis for regional ecological planning.

Comorbidity Patterns in Patients with Atopic Dermatitis Using Network Analysis in the EpiChron Study

Background: Atopic dermatitis (AD) is associated with different comorbidities. Methods: Retrospective, observational study based on clinical information from the individuals of the EpiChron Cohort Study (Aragon, Spain) with a diagnosis of AD between 1 January 2010 and 31 December 2018. We calculated the tetrachoric correlations of each pair of comorbidities to analyze the weight of the association between them. We used a cut-off point for statistical significance of p-value < 0.01. Results: The prevalence of AD in the EpiChron Cohort was 3.83%. The most frequently found comorbidities were respiratory, cardio-metabolic, cardiovascular, and mental health disorders. Comorbidities were combined into 17 disease patterns (15 in men and 11 in women), with some sex and age specificities. An infectious respiratory pattern was the most consistently described pattern across all ages and sexes, followed by a cardiometabolic pattern that appeared in patients over 18 years of age. Conclusions: Our study revealed the presence of different clinically meaningful comorbidity patterns in patients with AD. Our results can help to identify which comorbidities deserve special attention in these types of patients and to better understand the physio-pathological mechanisms underlying the disease associations identified. Further studies are encouraged to validate the results obtained in different clinical settings and populations.

Mapping nonlocal relationships between metadata and network structure with metadata-dependent encoding of random walks

Integrating structural information and metadata, such as gender, social status, or interests, enriches networks and enables a better understanding of the large-scale structure of complex systems. However, existing approaches to augment networks with metadata for community detection only consider immediately adjacent nodes and cannot exploit the nonlocal relationships between metadata and large-scale network structure present in many spatial and social systems. Here, we develop a flow-based community detection framework based on the map equation that integrates network information and metadata of distant nodes and reveals more complex relationships. We analyze social and spatial networks and find that our methodology can detect functional metadata-informed communities distinct from those derived solely from network information or metadata. For example, in a mobility network of London, we identify communities that reflect the heterogeneity of income distribution, and in a European power grid network, we identify communities that capture relationships between geography and energy prices beyond country borders.

Classification of multiple sclerosis clinical profiles using machine learning and grey matter connectome

Purpose: The main goal of this study is to investigate the discrimination power of Grey Matter (GM) thickness connectome data between Multiple Sclerosis (MS) clinical profiles using statistical and Machine Learning (ML) methods. Materials and Methods: A dataset composed of 90 MS patients acquired at the MS clinic of Lyon Neurological Hospital was used for the analysis. Four MS profiles were considered, corresponding to Clinical Isolated Syndrome (CIS), Relapsing-Remitting MS (RRMS), Secondary Progressive MS (SPMS), and Primary Progressive MS (PPMS). Each patient was classified in one of these profiles by our neurologist and underwent longitudinal MRI examinations including T1-weighted image acquisition at each examination, from which the GM tissue was segmented and the cortical GM thickness measured. Following the GM parcellation using two different atlases (FSAverage and Glasser 2016), the morphological connectome was built and six global metrics (Betweenness Centrality (BC), Assortativity (r), Transitivity (T), Efficiency (E g ), Modularity (Q) and Density (D)) were extracted. Based on their connectivity metrics, MS profiles were first statistically compared and second, classified using four different learning machines (Logistic Regression, Random Forest, Support Vector Machine and AdaBoost), combined in a higher level ensemble model by majority voting. Finally, the impact of the GM spatial resolution on the MS clinical profiles classification was analyzed. Results: Using binary comparisons between the four MS clinical profiles, statistical differences and classification performances higher than 0.7 were observed. Good performances were obtained when comparing the two early clinical forms, RRMS and PPMS (F1 score of 0.86), and the two neurodegenerative profiles, PPMS and SPMS (F1 score of 0.72). When comparing the two atlases, slightly better performances were obtained with the Glasser 2016 atlas, especially between RRMS with PPMS (F1 score of 0.83), compared to the FSAverage atlas (F1 score of 0.69). Also, the thresholding value for graph binarization was investigated suggesting more informative graph properties in the percentile range between 0.6 and 0.8. Conclusion: An automated pipeline was proposed for the classification of MS clinical profiles using six global graph metrics extracted from the GM morphological connectome of MS patients. This work demonstrated that GM morphological connectivity data could provide good classification performances by combining four simple ML models, without the cost of long and complex MR techniques, such as MR diffusion, and/or deep learning architectures.

Global brain network modularity dynamics after local optic nerve damage following noninvasive brain stimulation: an EEG-tracking study

Tightly connected clusters of nodes, called communities, interact in a time-dependent manner in brain functional connectivity networks (FCN) to support complex cognitive functions. However, little is known if and how different nodes synchronize their neural interactions to form functional communities ("modules") during visual processing and if and how this modularity changes postlesion (progression or recovery) following neuromodulation. Using the damaged optic nerve as a paradigm, we now studied brain FCN modularity dynamics to better understand module interactions and dynamic reconfigurations before and after neuromodulation with noninvasive repetitive transorbital alternating current stimulation (rtACS). We found that in both patients and controls, local intermodule interactions correlated with visual performance. However, patients' recovery of vision after treatment with rtACS was associated with improved interaction strength of pathways linked to the attention module, and it improved global modularity and increased the stability of FCN. Our results show that temporal coordination of multiple cortical modules and intermodule interaction are functionally relevant for visual processing. This modularity can be neuromodulated with tACS, which induces a more optimal balanced and stable multilayer modular structure for visual processing by enhancing the interaction of neural pathways with the attention network module.

Mapping the Nicotinic Acetylcholine Receptor Nanocluster Topography at the Cell Membrane with STED and STORM Nanoscopies

The cell-surface topography and density of nicotinic acetylcholine receptors (nAChRs) play a key functional role in the synapse. Here we employ in parallel two labeling and two super-resolution microscopy strategies to characterize the distribution of this receptor at the plasma membrane of the mammalian clonal cell line CHO-K1/A5. Cells were interrogated with two targeted techniques (confocal microscopy and stimulated emission depletion (STED) nanoscopy) and single-molecule nanoscopy (stochastic optical reconstruction microscopy, STORM) using the same fluorophore, Alexa Fluor 647, tagged onto either α-bungarotoxin (BTX) or the monoclonal antibody mAb35. Analysis of the topography of nanometer-sized aggregates (“nanoclusters”) was carried out using STORMGraph, a quantitative clustering analysis for single-molecule localization microscopy based on graph theory and community detection, and ASTRICS, an inter-cluster similarity algorithm based on computational geometry. Antibody-induced crosslinking of receptors resulted in nanoclusters with a larger number of receptor molecules and higher densities than those observed in BTX-labeled samples. STORM and STED provided complementary information, STED rendering a direct map of the mesoscale nAChR distribution at distances ~10-times larger than the nanocluster centroid distances measured in STORM samples. By applying photon threshold filtering analysis, we show that it is also possible to detect the mesoscale organization in STORM images.

In Alzheimer-prone brain regions, metabolism and risk-gene expression are strongly correlated

Neuroimaging in the preclinical phase of Alzheimer’s disease provides information crucial to early intervention, particularly in people with a high genetic risk. Metabolic network modularity, recently applied to the study of dementia, is increased in Alzheimer’s disease patients compared with controls, but network modularity in cognitively unimpaired elderly with various risks of developing Alzheimer’s disease needs to be determined. Based on their 5-year cognitive progression, we stratified 117 cognitively normal participants (78.3 ± 4.0 years of age, 52 women) into three age-matched groups, each with a different level of risk for Alzheimer’s disease. From their fluorodeoxyglucose PET we constructed metabolic networks, evaluated their modular structures using the Louvain algorithm, and compared them between risk groups. As the risk for Alzheimer’s disease increased, the metabolic connections among brain regions weakened and became more modular, indicating network fragmentation and functional impairment of the brain. We then set out to determine the correlation between regional brain metabolism, particularly in the modules derived from the previous analysis, and the regional expression of Alzheimer-risk genes in the brain, obtained from the Allen Human Brain Atlas. In all risk groups of this elderly population, the regional brain expression of most Alzheimer-risk genes showed a strong correlation with brain metabolism, particularly in the module that corresponded to regions of the brain that are affected earliest and most severely in Alzheimer’s disease. Among the genes, APOE and CD33 showed the strongest negative correlation and SORL1 showed the strongest positive correlation with brain metabolism. The Pearson correlation coefficients remained significant when contrasted against a null-hypothesis distribution of correlation coefficients across the whole transcriptome of 20 736 genes (SORL1: P = 0.0130; CD33, P = 0.0136; APOE: P = 0.0093). The strong regional correlation between Alzheimer-related gene expression in the brain and brain metabolism in older adults highlights the role of brain metabolism in the genesis of dementia.

Clustering of Cancer Attributed Networks by Dynamically and Jointly Factorizing Multi-Layer Graphs

The accumulated omic data provides an opportunity to exploit the mechanisms of cancers and poses a challenge for their integrative analysis. Although extensive efforts have been devoted to address this issue, the current algorithms result in undesirable performance because of the complexity of patterns and heterogeneity of data. In this study, the ultimate goal is to propose an effective and efficient algorithm (called NMF-DEC) to identify clusters by integrating the interactome and transcriptome data. By treating the expression profiles of genes as attributes of vertices in the gene interaction networks, we transform the integrative analysis of omic data into clustering of attributed networks. To circumvent the heterogeneity, we construct a similarity network for the attributes of genes and cast it into the common module detection problem in multi-layer networks. The NMF-DEC explores the relation between attributes and topological structure of networks by jointly factorizing the similarity and interaction networks with the same basis. In this optimization, the interaction network is dynamically updated and the information of attributes is dynamically incorporated, providing a better strategy to characterize the structure of modules in attributed networks. Extensive experiments indicate that compared with state-of-the-art baselines, NMF-DEC is more accurate on social network, and show better performance on cancer attributed networks, implying the superiority of the proposed methods for the integrative analysis of omic data.

Contact Tracing Research: A Literature Review Based on Scientific Collaboration Network

Contact tracing is a monitoring process including contact identification, listing, and follow-up, which is a key to slowing down pandemics of infectious diseases, such as COVID-19. In this study, we use the scientific collaboration network technique to explore the evolving history and scientific collaboration patterns of contact tracing. It is observed that the number of articles on the subject remained at a low level before 2020, probably because the practical significance of the contact tracing model was not widely accepted by the academic community. The COVID-19 pandemic has brought an unprecedented research boom to contact tracing, as evidenced by the explosion of the literature after 2020. Tuberculosis, HIV, and other sexually transmitted diseases were common types of diseases studied in contact tracing before 2020. In contrast, research on contact tracing regarding COVID-19 occupies a significantly large proportion after 2000. It is also found from the collaboration networks that academic teams in the field tend to conduct independent research, rather than cross-team collaboration, which is not conducive to knowledge dissemination and information flow.

Brokering or Sitting Between Two Chairs? A Group Perspective on Workplace Gossip

Brokerage is a central concept in the organization literature. It has been argued that individuals in broker positions—i.e., connecting otherwise disconnected parts within a firm’s social network—can control the flow of information. It would imply their increased relevance in workplace gossip. This allegation, however, has not been addressed empirically yet. To fill this gap, we apply social network analysis techniques to relational data from six organizations in Hungary. First, we identify informal groups and individuals in broker positions. Then, we use this information to predict the likelihood with which positive or negative gossip is reported. We find more gossip when the sender and receiver are part of the same group and more positive gossip about in-group rather than out-group targets. Individuals in broker positions are more likely the senders and targets of negative gossip. Finally, even if both the brokers and the boss(es) are the targets of their colleagues’ negative gossip, the combination of the two categories (bosses in broker positions) does not predict more negative gossip anymore. Results are discussed in relation to the theoretical accounts on brokerage that emphasize its power for information control but fail to recognize the pitfalls of being in such positions.

Link Pruning for Community Detection in Social Networks

Attempts to discover knowledge through data are gradually becoming diversified to understand complex aspects of social phenomena. Graph data analysis, which models and analyzes complex data as graphs, draws much attention as it combines the latest machine learning techniques. In this paper, we propose a new framework called link pruning for detecting clusters in complex networks, which leverages the cohesiveness of local structures by removing unimportant connections. Link pruning is a flexible framework that reduces the clustering problem in a highly mixed community structure to a simpler problem with a lowly mixed community structure. We analyze which similarities and curvatures defined on the pairs of nodes, which we call the link attributes, allow links inside and outside the community to have a different range of values. Using the link attributes, we design and analyze an algorithm that eliminates links with low attribute values to find a better community structure on the transformed graph with low mixing. Through extensive experiments, we have shown that clustering algorithms with link pruning achieve higher quality than existing algorithms in both synthetic and real-world social networks.

Impact of the COVID-19 pandemic on urban human mobility - A multiscale geospatial network analysis using New York bike-sharing data

The COVID-19 pandemic breaking out at the end of 2019 has seriously impacted urban human mobility and poses great challenges for traffic management and urban planning. An understanding of this influence from multiple perspectives is urgently needed. In this study, we propose a multiscale geospatial network framework for the analysis of bike-sharing data, aiming to provide a new perspective for the exploration of the pandemic impact on urban human mobility. More specifically, we organize the bike-sharing data into a network representation, and divide the network into a three-scale structure, ranging from the whole bike system at the macroscale, to the network community at the mesoscale and then to the bicycle station at the microscale. The spatiotemporal analysis of bike-sharing data at each scale is combined with visualization methods for an intuitive understanding of the patterns. We select New York City, one of the most seriously influenced city by the pandemic, as the study area, and used Citi Bike bike-sharing data from January to April in 2019 and 2020 in this area for the investigation. The analysis results show that with the development of the pandemic, the riding flow and its spatiotemporal distribution pattern changed significantly, which had a series of effects on the use and management of bikes in the city. These findings may provide useful references during the pandemic for various stakeholders, e.g., citizens for their travel planning, bike-sharing companies for bicycle dispatching and bicycle disinfection management, and governments for traffic management.

A Graph-Cut-Based Approach to Community Detection in Networks

Networks can be used to model various aspects of our lives as well as relations among many real-world entities and objects. To detect a community structure in a network can enhance our understanding of the characteristics, properties, and inner workings of the network. Therefore, there has been significant research on detecting and evaluating community structures in networks. Many fields, including social sciences, biology, engineering, computer science, and applied mathematics, have developed various methods for analyzing and detecting community structures in networks. In this paper, a new community detection algorithm, which repeats the process of dividing a community into two smaller communities by finding a minimum cut, is proposed. The proposed algorithm is applied to some example network data and shows fairly good community detection results with comparable modularity Q values.

Collusion-Resistant Worker Set Selection for Transparent and Verifiable Voting

Collusion occurs when multiple malicious participants of a distributed protocol work together to sabotage or spy on honest participants. decentralized protocols often rely on a subset of participants called workers for critical operations. Collusion between workers can be particularly harmful to the security of the protocol. We propose two protocols that select a subset of workers from the set of participants such that the probability of the workers colluding together is minimized. Our first solution is a decentralized protocol that randomly selects workers in a verifiable manner without any trusted entities. The second solution is an algorithm that uses a social graph of participants and community detection to select workers that are socially distant in order to reduce the probability of collusion. We present our solutions in the context of a decentralized voting protocol proposed by Schiedermeier et al. (A transparent referendum protocol with immutable proceedings and verifiable outcome for trustless networks, Springer, Cham, 2019) that guarantees transparency and verifiability. Enabling collusion-resistance in order to ensure democratic voting is clearly of paramount importance thus the voting protocol provides a suitable use case for our solutions.

Tumor-induced double positive T cells display distinct lineage commitment mechanisms and functions

Transcription factors ThPOK and Runx3 regulate the differentiation of "helper" CD4+ and "cytotoxic" CD8+ T cell lineages respectively, inducing single positive (SP) T cells that enter the periphery with the expression of either the CD4 or CD8 co-receptor. Despite the expectation that these cell fates are mutually exclusive and that mature CD4+CD8+ double positive (DP) T cells are present in healthy individuals and augmented in the context of disease, yet their molecular features and pathophysiologic role are disputed. Here, we show DP T cells in murine and human tumors as a heterogenous population originating from SP T cells which re-express the opposite co-receptor and acquire features of the opposite cell type's phenotype and function following TCR stimulation. We identified distinct clonally expanded DP T cells in human melanoma and lung cancer by scRNA sequencing and demonstrated their tumor reactivity in cytotoxicity assays. Our findings indicate that antigen stimulation induces SP T cells to differentiate into DP T cell subsets gaining in polyfunctional characteristics.

Research Hotspots and Trends Analysis of Patellar Instability: A Bibliometric Analysis from 2001 to 2021

Background

Patellar instability is a common multifactorial disease in orthopedics, which seriously affects the quality of life. Because of the unified pathogeny, diagnosis and treatment, patellar instability has gradually attracted the interest of more scholars these years, resulting in an explosive growth in the research output. This study aims to summarize the knowledge structure and development trend in the field from the perspective of bibliometrics.

Methods

The data of articles and reviews on patellar instability was extracted from the Web of Science database. The Microsoft Excel, R-bibliometrix, CiteSpace, VOSviewer, Pajek software are comprehensively used to scientifically analyze the data quantitatively and qualitatively.

Results

Totally, 2,155 papers were identified, mainly from North America, Western Europe and East Asia. Until December 31, 2021, the United States has contributed the most articles (1,828) and the highest total citations (17,931). Hospital for Special Surgery and professor Andrew A Amis are the most prolific institutions and the most influential authors respectively. Through the analysis of citations and keywords based on a large number of literatures, “medial patellofemoral ligament construction”, “tibial tubercle-trochlear groove (TT-TG) distance”, “epidemiological prevalence”, “multifactor analysis of etiology, clinical outcome and radiographic landmarks “ were identified to be the most promising research directions.

Conclusions

This is the first bibliometric study to comprehensively summarize the research trend and development of patellar instability. The result of our research provides the updated perspective for scholars to understand the key information in this field, and promote future research to a great extent.

Probabilistically Weighted Multilayer Networks disclose the link between default mode network instability and psychosis-like experiences in healthy adults

The brain is a complex system in which the functional interactions among its subunits vary over time. The trajectories of this dynamic variation contribute to inter-individual behavioral differences and psychopathologic phenotypes. Despite many methodological advancements, the study of dynamic brain networks still relies on biased assumptions in the temporal domain. The current paper has two goals. First, we present a novel method to study multilayer networks: by modelling intra-nodal connections in a probabilistic, biologically driven way, we introduce a temporal resolution of the multilayer network based on signal similarity across time series. This new method is tested on synthetic networks by varying the number of modules and the sources of noise in the simulation. Secondly, we implement these probabilistically weighted (PW) multilayer networks to study the association between network dynamics and subclinical, psychosis-relevant personality traits in healthy adults. We show that the PW method for multilayer networks outperforms the standard procedure in modular detection and is less affected by increasing noise levels. Additionally, the PW method highlighted associations between the temporal instability of default mode network connections and psychosis-like experiences in healthy adults. PW multilayer networks allow an unbiased study of dynamic brain functioning and its behavioral correlates.

PISA data clusters reveal student and school inequality that affects results

The data from the PISA survey show that student performance correlates with socio-economic background, that private schools have higher results and more privileged students, and that this varies between countries. We explore this further and analyze the PISA data using methods from network theory and find clusters of countries whose students have similar performance and socio-economic background. Interestingly, we find a cluster of countries, including China, Spain and Portugal, characterized by less privileged students performing well. When considering private schools only, some countries, such as Portugal and Brazil, are in a cluster with mostly wealthy countries characterized by privileged students. Swedish grades are compared to PISA results, and we see that the higher grades in private schools are in line with the PISA results, suggesting that there is no grade inflation in this case, but differences in socio-economic background suggest that this is due to school segregation.

Benefits and Challenges of Pre-clustered Network-Based Pathway Analysis

Functional analysis of gene sets derived from experiments is typically done by pathway annotation. Although many algorithms exist for analyzing the association between a gene set and a pathway, an issue which is generally ignored is that gene sets often represent multiple pathways. In such cases an association to a pathway is weakened by the presence of genes associated with other pathways. A way to counteract this is to cluster the gene set into more homogenous parts before performing pathway analysis on each module. We explored whether network-based pre-clustering of a query gene set can improve pathway analysis. The methods MCL, Infomap, and MGclus were used to cluster the gene set projected onto the FunCoup network. We characterized how well these methods are able to detect individual pathways in multi-pathway gene sets, and applied each of the clustering methods in combination with four pathway analysis methods: Gene Enrichment Analysis, BinoX, NEAT, and ANUBIX. Using benchmarks constructed from the KEGG pathway database we found that clustering can be beneficial by increasing the sensitivity of pathway analysis methods and by providing deeper insights of biological mechanisms related to the phenotype under study. However, keeping a high specificity is a challenge. For ANUBIX, clustering caused a minor loss of specificity, while for BinoX and NEAT it caused an unacceptable loss of specificity. GEA had very low sensitivity both before and after clustering. The choice of clustering method only had a minor effect on the results. We show examples of this approach and conclude that clustering can improve overall pathway annotation performance, but should only be used if the used enrichment method has a low false positive rate.

A global Anopheles gambiae gene co-expression network constructed from hundreds of experimental conditions with missing values

BACKGROUND

Gene co-expression networks (GCNs) can be used to determine gene regulation and attribute gene function to biological processes. Different high throughput technologies, including one and two-channel microarrays and RNA-sequencing, allow evaluating thousands of gene expression data simultaneously, but these methodologies provide results that cannot be directly compared. Thus, it is complex to analyze co-expression relations between genes, especially when there are missing values arising for experimental reasons. Networks are a helpful tool for studying gene co-expression, where nodes represent genes and edges represent co-expression of pairs of genes.

RESULTS

In this paper, we establish a method for constructing a gene co-expression network for the Anopheles gambiae transcriptome from 257 unique studies obtained with different methodologies and experimental designs. We introduce the sliding threshold approach to select node pairs with high Pearson correlation coefficients. The resulting network, which we name AgGCN1.0, is robust to random removal of conditions and has similar characteristics to small-world and scale-free networks. Analysis of network sub-graphs revealed that the core is largely comprised of genes that encode components of the mitochondrial respiratory chain and the ribosome, while different communities are enriched for genes involved in distinct biological processes.

CONCLUSION

Analysis of the network reveals that both the architecture of the core sub-network and the network communities are based on gene function, supporting the power of the proposed method for GCN construction. Application of network science methodology reveals that the overall network structure is driven to maximize the integration of essential cellular functions, possibly allowing the flexibility to add novel functions.

Human-supervised clustering of multidimensional data using crowdsourcing

Clustering is a central task in many data analysis applications. However, there is no universally accepted metric to decide the occurrence of clusters. Ultimately, we have to resort to a consensus between experts. The problem is amplified with high-dimensional datasets where classical distances become uninformative and the ability of humans to fully apprehend the distribution of the data is challenged. In this paper, we design a mobile human-computing game as a tool to query human perception for the multidimensional data clustering problem. We propose two clustering algorithms that partially or entirely rely on aggregated human answers and report the results of two experiments conducted on synthetic and real-world datasets. We show that our methods perform on par or better than the most popular automated clustering algorithms. Our results suggest that hybrid systems leveraging annotations of partial datasets collected through crowdsourcing platforms can be an efficient strategy to capture the collective wisdom for solving abstract computational problems.

Evaluating Methods for Efficient Community Detection in Social Networks

Exploring a community is an important aspect of social network analysis because it can be seen as a crucial way to decompose specific graphs into smaller graphs based on interactions between users. The process of discovering common features between groups of users, entitled “community detection”, is a fundamental feature for social network analysis, wherein the vertices represent the users and the edges their relationships. Our study focuses on identifying such phenomena on the Twitter graph of posts and on determining communities, which contain users with similar features. This paper presents the evaluation of six established community-discovery algorithms, namely Breadth-First Search, CNM, Louvain, MaxToMin, Newman–Girvan and Propinquity Dynamics, in terms of four widely used graphs and a collection of data fetched from Twitter about man-made and physical data. Furthermore, the size of each community, expressed as a percentage of the total number of vertices, is identified for the six particular algorithms, and corresponding results are extracted. In terms of user-based evaluation, we indicated to some students the communities that were extracted by every algorithm, with a corresponding user and their tweets in the grouping and considered three different alternatives for the extracted communities: “dense community”, “sparse community” and “in-between”. Our findings suggest that the community-detection algorithms can assist in identifying dense group of users.

Multimorbidity Profiles and Infection Severity in COVID-19 Population Using Network Analysis in the Andalusian Health Population Database

Identifying the population at risk of COVID-19 infection severity is a priority for clinicians and health systems. Most studies to date have only focused on the effect of specific disorders on infection severity, without considering that patients usually present multiple chronic diseases and that these conditions tend to group together in the form of multimorbidity patterns. In this large-scale epidemiological study, including primary and hospital care information of 166,242 patients with confirmed COVID-19 infection from the Spanish region of Andalusia, we applied network analysis to identify multimorbidity profiles and analyze their impact on the risk of hospitalization and mortality. Our results showed that multimorbidity was a risk factor for COVID-19 severity and that this risk increased with the morbidity burden. Individuals with advanced cardio-metabolic profiles frequently presented the highest infection severity risk in both sexes. The pattern with the highest severity associated in men was present in almost 28.7% of those aged ≥ 80 years and included associations between cardiovascular, respiratory, and metabolic diseases; age-adjusted odds ratio (OR) 95% confidence interval (1.71 (1.44–2.02)). In women, similar patterns were also associated the most with infection severity, in 7% of 65–79-year-olds (1.44 (1.34–1.54)) and in 29% of ≥80-year-olds (1.35 (1.18–1.53)). Patients with mental health patterns also showed one of the highest risks of COVID-19 severity, especially in women. These findings strongly recommend the implementation of personalized approaches to patients with multimorbidity and SARS-CoV-2 infection, especially in the population with high morbidity burden.

HiCBin: binning metagenomic contigs and recovering metagenome-assembled genomes using Hi-C contact maps

Recovering high-quality metagenome-assembled genomes (MAGs) from complex microbial ecosystems remains challenging. Recently, high-throughput chromosome conformation capture (Hi-C) has been applied to simultaneously study multiple genomes in natural microbial communities. We develop HiCBin, a novel open-source pipeline, to resolve high-quality MAGs utilizing Hi-C contact maps. HiCBin employs the HiCzin normalization method and the Leiden clustering algorithm and includes the spurious contact detection into binning pipelines for the first time. HiCBin is validated on one synthetic and two real metagenomic samples and is shown to outperform the existing Hi-C-based binning methods. HiCBin is available at https://github.com/dyxstat/HiCBin .

Identifying multimorbidity profiles associated with COVID-19 severity in chronic patients using network analysis in the PRECOVID Study

A major risk factor of COVID-19 severity is the patient's health status at the time of the infection. Numerous studies focused on specific chronic diseases and identified conditions, mainly cardiovascular ones, associated with poor prognosis. However, chronic diseases tend to cluster into patterns, each with its particular repercussions on the clinical outcome of infected patients. Network analysis in our population revealed that not all cardiovascular patterns have the same risk of COVID-19 hospitalization or mortality and that this risk depends on the pattern of multimorbidity, besides age and sex. We evidenced that negative outcomes were strongly related to patterns in which diabetes and obesity stood out in older women and men, respectively. In younger adults, anxiety was another disease that increased the risk of severity, most notably when combined with menstrual disorders in women or atopic dermatitis in men. These results have relevant implications for organizational, preventive, and clinical actions to help meet the needs of COVID-19 patients.

Do Informational and Emotional Elements Differ between Online Psychological and Physiological Disease Communities in China? A Comparative Study of Depression and Diabetes

Disease-specific online health communities provide a convenient and common platform for patients to share experiences, change information, provide and receive social support. This study aimed to compare differences between online psychological and physiological disease communities in topics, sentiment, participation, and emotional contagion patterns using multiple methods as well as to discuss how to satisfy the users' different informational and emotional needs. We chose the online depression and diabetes communities on the Baidu Tieba platform as the data source. Topic modeling and theme coding were employed to analyze discussion preferences for various topic categories. Sentiment analysis was used to identify the sentiment polarity of each post and comment. The social network was used to represent the users' interaction and emotional flows to discover the differences in participation and emotional contagion patterns between psychological and physiological disease communities. The results revealed that people affected by depression focused more on their symptoms and social relationships, while people affected by diabetes were more likely to discuss treatment and self-management behavior. In the depression community, there were obvious interveners spreading positive emotions and more core users in the negative emotional contagion network. In the diabetes community, emotional contagion was less prevalent and core users in positive and negative emotional contagion networks were basically the same. The study reveals insights into the differences between online psychological and physiological disease communities, providing a greater understanding of the users' informational and emotional needs expressed online. These results are helpful for society to provide actual medical assistance and deploy health interventions based on disease types.

Bike-Sharing Demand Prediction at Community Level under COVID-19 Using Deep Learning

An important question in planning and designing bike-sharing services is to support the user’s travel demand by allocating bikes at the stations in an efficient and reliable manner which may require accurate short-time demand prediction. This study focuses on the short-term forecasting, 15 min ahead, of the shared bikes demand in Montreal using a deep learning approach. Having a set of bike trips, the study first identifies 6 communities in the bike-sharing network using the Louvain algorithm. Then, four groups of LSTM-based architectures are adopted to predict pickup demand in each community. A univariate ARIMA model is also used to compare results as a benchmark. The historical trip data from 2017 to 2021 are used in addition to the extra inputs of demand related engineered features, weather conditions, and temporal variables. The selected timespan allows predicting bike demand during the COVID-19 pandemic. Results show that the deep learning models significantly outperform the ARIMA one. The hybrid CNN-LSTM achieves the highest prediction accuracy. Furthermore, adding the extra variables improves the model performance regardless of its architecture. Thus, using the hybrid structure enriched with additional input features provides a better insight into the bike demand patterns, in support of bike-sharing operational management.

Network Structure and Community Evolution Online: Behavioral and Emotional Changes in Response to COVID-19

Background: The measurement and identification of changes in the social structure in response to an exceptional event like COVID-19 can facilitate a more informed public response to the pandemic and provide fundamental insights on how collective social processes respond to extreme events. Objective: In this study, we built a generalized framework for applying social media data to understand public behavioral and emotional changes in response to COVID-19. Methods: Utilizing a complete dataset of Sina Weibo posts published by users in Wuhan from December 2019 to March 2020, we constructed a time-varying social network of 3.5 million users. In combination with community detection, text analysis, and sentiment analysis, we comprehensively analyzed the evolution of the social network structure, as well as the behavioral and emotional changes across four main stages of Wuhan's experience with the epidemic. Results: The empirical results indicate that almost all network indicators related to the network's size and the frequency of social interactions increased during the outbreak. The number of unique recipients, average degree, and transitivity increased by 24, 23, and 19% during the severe stage than before the outbreak, respectively. Additionally, the similarity of topics discussed on Weibo increased during the local peak of the epidemic. Most people began discussing the epidemic instead of the more varied cultural topics that dominated early conversations. The number of communities focused on COVID-19 increased by nearly 40 percent of the total number of communities. Finally, we find a statistically significant "rebound effect" by exploring the emotional content of the users' posts through paired sample t-test (P = 0.003). Conclusions: Following the evolution of the network and community structure can explain how collective social processes changed during the pandemic. These results can provide data-driven insights into the development of public attention during extreme events.

Synwalk: community detection via random walk modelling

Complex systems, abstractly represented as networks, are ubiquitous in everyday life. Analyzing and understanding these systems requires, among others, tools for community detection. As no single best community detection algorithm can exist, robustness across a wide variety of problem settings is desirable. In this work, we present Synwalk, a random walk-based community detection method. Synwalk builds upon a solid theoretical basis and detects communities by synthesizing the random walk induced by the given network from a class of candidate random walks. We thoroughly validate the effectiveness of our approach on synthetic and empirical networks, respectively, and compare Synwalk’s performance with the performance of Infomap and Walktrap (also random walk-based), Louvain (based on modularity maximization) and stochastic block model inference. Our results indicate that Synwalk performs robustly on networks with varying mixing parameters and degree distributions. We outperform Infomap on networks with high mixing parameter, and Infomap and Walktrap on networks with many small communities and low average degree. Our work has a potential to inspire further development of community detection via synthesis of random walks and we provide concrete ideas for future research.

OUP accepted manuscript

In the last years, many studies were able to identify associations between common genetic variants and complex diseases. However, the mechanistic biological links explaining these associations are still mostly unknown. Common variants are usually associated with a relatively small effect size, suggesting that interactions among multiple variants might be a major genetic component of complex diseases. Hence, elucidating the presence of functional relations among variants may be fundamental to identify putative variants’ interactions. To this aim, we developed Polympact, a web-based resource that allows to explore functional relations among human common variants by exploiting variants’ functional element landscape, their impact on transcription factor binding motifs, and their effect on transcript levels of protein-coding genes. Polympact characterizes over 18 million common variants and allows to explore putative relations by combining clustering analysis and innovative similarity and interaction network models. The properties of the network models were studied and the utility of Polympact was demonstrated by analysing the rich sets of Breast Cancer and Alzheimer's GWAS variants. We identified relations among multiple variants, suggesting putative interactions. Polympact is freely available at bcglab.cibio.unitn.it/polympact.

Adapting k-means for graph clustering

We propose two new algorithms for clustering graphs and networks. The first, called K‑algorithm, is derived directly from the k-means algorithm. It applies similar iterative local optimization but without the need to calculate the means. It inherits the properties of k-means clustering in terms of both good local optimization capability and the tendency to get stuck at a local optimum. The second algorithm, called the M-algorithm, gradually improves on the results of the K-algorithm to find new and potentially better local optima. It repeatedly merges and splits random clusters and tunes the results with the K-algorithm. Both algorithms are general in the sense that they can be used with different cost functions. We consider the conductance cost function and also introduce two new cost functions, called inverse internal weight and mean internal weight. According to our experiments, the M-algorithm outperforms eight other state-of-the-art methods. We also perform a case study by analyzing clustering results of a disease co-occurrence network, which demonstrate the usefulness of the algorithms in an important real-life application.

A computational exploration of resilience and evolvability of protein-protein interaction networks

Protein-protein interaction (PPI) networks represent complex intra-cellular protein interactions, and the presence or absence of such interactions can lead to biological changes in an organism. Recent network-based approaches have shown that a phenotype's PPI network's resilience to environmental perturbations is related to its placement in the tree of life; though we still do not know how or why certain intra-cellular factors can bring about this resilience. Here, we explore the influence of gene expression and network properties on PPI networks' resilience. We use publicly available data of PPIs for E. coli, S. cerevisiae, and H. sapiens, where we compute changes in network resilience as new nodes (proteins) are added to the networks under three node addition mechanisms-random, degree-based, and gene-expression-based attachments. By calculating the resilience of the resulting networks, we estimate the effectiveness of these node addition mechanisms. We demonstrate that adding nodes with gene-expression-based preferential attachment (as opposed to random or degree-based) preserves and can increase the original resilience of PPI network in all three species, regardless of gene expression distribution or network structure. These findings introduce a general notion of prospective resilience, which highlights the key role of network structures in understanding the evolvability of phenotypic traits.

Modularity maximization as a flexible and generic framework for brain network exploratory analysis

The modular structure of brain networks supports specialized information processing, complex dynamics, and cost-efficient spatial embedding. Inter-individual variation in modular structure has been linked to differences in performance, disease, and development. There exist many data-driven methods for detecting and comparing modular structure, the most popular of which is modularity maximization. Although modularity maximization is a general framework that can be modified and reparamaterized to address domain-specific research questions, its application to neuroscientific datasets has, thus far, been narrow. Here, we highlight several strategies in which the "out-of-the-box" version of modularity maximization can be extended to address questions specific to neuroscience. First, we present approaches for detecting "space-independent" modules and for applying modularity maximization to signed matrices. Next, we show that the modularity maximization frame is well-suited for detecting task- and condition-specific modules. Finally, we highlight the role of multi-layer models in detecting and tracking modules across time, tasks, subjects, and modalities. In summary, modularity maximization is a flexible and general framework that can be adapted to detect modular structure resulting from a wide range of hypotheses. This article highlights multiple frontiers for future research and applications.

Community detection in large scale congested urban road networks

Traffic congestion in large urban networks may take different shapes and propagates non-uniformly variations from day to day. Given the fact that congestion on a road segment is spatially correlated to adjacent roads and propagates spatiotemporally with finite speed, it is essential to describe the main pockets of congestion in a city with a small number of clusters. For example, the perimeter control with macroscopic fundamental diagrams is one of the effective traffic management tools. Perimeter control adjusts the inflow to pre-specified regions of a city through signal timing on the border of a region in order to optimize the traffic condition within the region. The precision of macroscopic fundamental diagrams depends on the homogeneity of traffic condition on road segments of the region. Hence, previous studies have defined the boundaries of the region under perimeter control subjected to the regional homogeneity. In this study, a cost-effective method is proposed for the mentioned problem that simultaneously considers homogeneity, contiguity and compactness of clusters and has a shorter computational time. Since it is necessary to control the cost and complexity of perimeter control in terms of the number of traffic signals, sparse parts of the network could be potential candidates for boundaries. Therefore, a community detection method (Infomap) is initially adopted and then those clusters are improved by refining the communities in relation to roads with the highest heterogeneity. The proposed method is applied to Shenzhen, China and San Francisco, USA and the outcomes are compared to previous studies. The results of comparison reveal that the proposed method is as effective as the best previous methods in detecting homogenous communities, but it outperforms them in contiguity. It is worth noting that this is the first method that guarantees the connectedness of clusters, which is a prerequisite of perimeter control.

Spatial variation in species' roles in host-helminth networks

Species interactions may vary considerably across space as a result of spatial and environmental gradients. With respect to host-parasite interactions, this suggests that host and parasite species may play different functional roles across the different networks they occur in. Using a global occurrence database of helminth parasites, we examine the conservation of species' roles using data on host-helminth interactions from 299 geopolitical locations. Defining species' roles in a two-dimensional space which captures the tendency of species to be more densely linked within species subgroups than between subgroups, we quantified species' roles in two ways, which captured if and which species' roles are conserved by treating species' utilization of this two-dimensional space as continuous, while also classifying species into categorical roles. Both approaches failed to detect the conservation of species' roles for a single species out of over 38 000 host and helminth parasite species. Together, our findings suggest that species' roles in host-helminth networks may not be conserved, pointing to the potential role of spatial and environmental gradients, as well as the importance of the context of the local host and helminth parasite community. This article is part of the theme issue 'Infectious disease macroecology: parasite diversity and dynamics across the globe'.