Similar Network Research Articles

Designing patient-oriented combination therapies for acute myeloid leukemia based on efficacy/toxicity integration and bipartite network modeling

Acute myeloid leukemia (AML), a heterogeneous and aggressive blood cancer, does not respond well to single-drug therapy. A combination of drugs is required to effectively treat this disease. Computational models are critical for combination therapy discovery due to the tens of thousands of two-drug combinations, even with approved drugs. While predicting synergistic drugs is the focus of current methods, few consider drug efficacy and potential toxicity, which are crucial for treatment success. To find effective new drug candidates, we constructed a bipartite network using patient-derived tumor samples and drugs. The network is based on drug-response screening and summarizes all treatment response heterogeneity as drug response weights. This bipartite network is then projected onto the drug part, resulting in the drug similarity network. Distinct drug clusters were identified using community detection methods, each targeting different biological processes and pathways as revealed by enrichment and pathway analysis of the drugs’ protein targets. Four drugs with the highest efficacy and lowest toxicity from each cluster were selected and tested for drug sensitivity using cell viability assays on various samples. Results show that ruxolitinib-ulixertinib and sapanisertib-LY3009120 are the most effective combinations with the least toxicity and the best synergistic effect on blast cells. These findings lay the foundation for personalized and successful AML therapies, ultimately leading to the development of drug combinations that can be used alongside standard first-line AML treatment.

Sequence similarity networks bear out hierarchical relationships of green cytochrome P450.

Land plants have diversified enzyme families. One of the most prominent is the cytochrome P450 (CYP or CYP450) family. With over 443,000 CYP proteins sequenced across the tree of life, CYPs are ubiquitous in archaea, bacteria, and eukaryotes. Here, we focused on land plants and algae to study the role of CYP diversification. CYPs, acting as monooxygenases, catalyze hydroxylation reactions crucial for specialized plant metabolic pathways, including detoxification and phytohormone production; the CYPome consists of one enormous superfamily that is divided into clans and families. Their evolutionary history speaks of high substrate promiscuity; radiation and functional diversification have yielded numerous CYP families. To understand the evolutionary relationships within the CYPs, we employed sequence similarity network analyses. We recovered distinct clusters representing different CYP families, reflecting their diversified sequences that we link to the prediction of functionalities. Hierarchical clustering and phylogenetic analysis further elucidated relationships between CYP clans, uncovering their shared deep evolutionary history. We explored the distribution and diversification of CYP subfamilies across plant and algal lineages, uncovering novel candidates and providing insights into the evolution of these enzyme families. This identified unexpected relationships between CYP families, such as the link between CYP82 and CYP74, shedding light on their roles in plant defense signaling pathways. Our approach provides a methodology that brings insights into the emergence of new functions within the CYP450 family, contributing to the evolutionary history of plants and algae. These insights can be further validated and implemented via experimental setups under various external conditions.

Chromatin accessibility profiling reveals that human fibroblasts respond to mechanical stimulation in a cell-specific manner.

Fibroblasts in the skin are highly heterogeneous, both in vivo and in vitro. One difference between follicular (dermal papilla fibroblasts [DP]) and interfollicular fibroblasts (papillary fibroblasts [PFi]) in vitro is their ability to differentiate in response to osteogenic media (OM), or mechanical stimulation. Here, we asked whether differences in the ability of DP and PFi to respond to differentiation stimuli are due to differences in chromatin accessibility. We performed chromatin accessibility and transcriptional profiling of DP and PFi in human skin, which arise from a common progenitor during development, yet display distinct characteristics in adult tissue and in vitro. We found that cells cultured in growth media had unique chromatin accessibility profiles; however, these profiles control similar functional networks. Upon introduction of a chemical perturbation (OM) to promote differentiation, we observed a divergence not only in the accessible chromatin signatures but also in the functional networks controlled by these signatures. The biggest divergence between DP and PFi was observed when we applied 2 perturbations to cells: growth in OM and mechanical stimulation (a shock wave [OMSW]). DP readily differentiate into bone in OMSW conditions, while PFi lack differentiation capability in vitro. In the DP we found a number of uniquely accessible promoters that controlled osteogenic interaction networks associated with bone and differentiation functions. Using ATAC-seq and RNA-seq we found that the combination of 2 stimuli (OMSW) could result in significant changes in chromatin accessibility associated with osteogenic differentiation, but only within the DP (capable of osteogenic differentiation). De novo motif analysis identified enrichment of motifs bound by the TEA domain (TEAD) family of transcription factors, and inter-cell comparisons (UpSet analysis) displayed large groups of genes to be unique to single cell types and conditions. Our results suggest that these 2 stimuli (OMSW) elicit cell-specific responses by modifying chromatin accessibility of osteogenic-related gene promoters.

Self-supervised few-shot learning for real-time traffic sign classification

Although supervised approaches for traffic sign classification have demonstrated excellent performance, they are limited to classifying several traffic signs defined in the training dataset. This prevents them from being applied to different domains, i.e., different countries. Herein, we propose a self-supervised approach for few-shot learning-based traffic sign classification. A center-awareness similarity network is designed for the traffic sign problem and trained using an optical flow dataset. Unlike existing supervised traffic sign classification methods, the proposed method does not depend on traffic sign categories defined by the training dataset. It applies to any traffic signs from different countries. We construct a Korean traffic sign classification (KTSC) dataset, including 6000 traffic sign samples and 59 categories. We evaluate the proposed method with baseline methods using the KTSC, German traffic sign, and Belgian traffic sign classification datasets. Experimental results show that the proposed method extends the ability of existing supervised methods and can classify any traffic sign, regardless of region/country dependence. Furthermore, the proposed approach significantly outperforms baseline methods for patch similarity. This approach provides a flexible and robust solution for classifying traffic signs, allowing for accurate categorization of every traffic sign, regardless of regional or national differences.

MCHAN: Prediction of Human Microbe-drug Associations Based on Multiview Contrastive Hypergraph Attention Network

Background: Complex and diverse microbial communities play a pivotal role in human health and have become a new drug target. Exploring the connections between drugs and microbes not only provides profound insights into their mechanisms but also drives progress in drug discovery and repurposing. The use of wet lab experiments to identify associations is time-consuming and laborious. Hence, the advancement of precise and efficient computational methods can effectively improve the efficiency of association identification between microorganisms and drugs. Objective: In this experiment, we propose a new deep learning model, a new multiview comparative hypergraph attention network (MCHAN) method for human microbe–drug association prediction. Methods: First, we fuse multiple similarity matrices to obtain a fused microbial and drug similarity network. By combining graph convolutional networks with attention mechanisms, we extract key information from multiple perspectives. Then, we construct two network topologies based on the above fused data. One topology incorporates the concept of hypernodes to capture implicit relationships between microbes and drugs using virtual nodes to construct a hyperheterogeneous graph. Next, we propose a cross-contrastive learning task that facilitates the simultaneous guidance of graph embeddings from both perspectives, without the need for any labels. This approach allows us to bring nodes with similar features and network topologies closer while pushing away other nodes. Finally, we employ attention mechanisms to merge the outputs of the GCN and predict the associations between drugs and microbes. Results: To confirm the effectiveness of this method, we conduct experiments on three distinct datasets. The results demonstrate that the MCHAN model surpasses other methods in terms of performance. Furthermore, case studies provide additional evidence confirming the consistent predictive accuracy of the MCHAN model. Conclusion: MCHAN is expected to become a valuable tool for predicting potential associations between microbiota and drugs in the future.

Predicting abiotic stress-responsive miRNA in plants based on multi-source features fusion and graph neural network

BackgroundMore and more studies show that miRNA plays a crucial role in plants' response to different abiotic stresses. However, traditional experimental methods are often expensive and inefficient, so it is important to develop efficient and economical computational methods. Although researchers have developed machine learning-based method, the information of miRNAs and abiotic stresses has not been fully exploited. Therefore, we propose a novel approach based on graph neural networks for predicting potential miRNA-abiotic stress associations.ResultsIn this study, we fully considered the multi-source feature information from miRNAs and abiotic stresses, and calculated and integrated the similarity network of miRNA and abiotic stress from different feature perspectives using multiple similarity measures. Then, the above multi-source similarity network and association information between miRNAs and abiotic stresses are effectively fused through heterogeneous networks. Subsequently, the Restart Random Walk (RWR) algorithm is employed to extract global structural information from heterogeneous networks, providing feature vectors for miRNA and abiotic stress. After that, we utilized the graph autoencoder based on GIN (Graph Isomorphism Networks) to learn and reconstruct a miRNA-abiotic stress association matrix to obtain potential miRNA-abiotic stress associations. The experimental results show that our model is superior to all known methods in predicting potential miRNA-abiotic stress associations, and the AUPR and AUC metrics of our model achieve 98.24% and 97.43%, respectively, under five-fold cross-validation.ConclusionsThe robustness and effectiveness of our proposed model position it as a valuable approach for advancing the field of miRNA-abiotic stress association prediction.

Biosynthesis of Cytidine Diphosphate-6-d-Glucitol for the Capsular Polysaccharides of Campylobacter jejuni.

Campylobacter jejuni is a Gram-negative pathogenic bacterium commonly found in chickens and is the leading cause of human diarrheal disease worldwide. The various serotypes of C.jejuni produce structurally distinct capsular polysaccharides (CPSs) on the exterior surfaces of the cell wall. The capsular polysaccharide from C.jejuni serotype HS:5 is composed of a repeating sequence of d-glycero-d-manno-heptose and d-glucitol-6-phosphate. We previously defined the pathway for the production of d-glycero-d-manno-heptose in C.jejuni. Here, we elucidate the biosynthetic pathway for the assembly of cytidine diphosphate (CDP)-6-d-glucitol by the combined action of two previously uncharacterized enzymes. The first enzyme catalyzes the formation of CDP-6-d-fructose from cytidine triphosphate (CTP) and d-fructose-6-phosphate. The second enzyme reduces CDP-6-d-fructose with NADPH to generate CDP-6-d-glucitol. Using sequence similarity network (SSN) and genome neighborhood network (GNN) analyses, we predict that these pairs of proteins are responsible for the biosynthesis of CDP-6-d-glucitol and/or CDP-d-mannitol in the lipopolysaccharides (LPSs) and capsular polysaccharides in more than 200 other organisms. In addition, high resolution X-ray structures of the second enzyme are reported, which provide novel insight into the manner in which an open-chain nucleotide-linked sugar is harbored in an active site cleft.

Alterations in the multilayer network in patients with rapid eye movement sleep behaviour disorder.

This study aimed to reveal the pathophysiology of isolated rapid eye movement sleep behaviour disorder (RBD) in patients using multilayer network analysis. Participants eligible for isolated RBD were included and verified via polysomnography. Both iRBD patients and healthy controls underwent brain MRI, including T1-weighted imaging and diffusion tensor imaging. Grey matter matrix was derived from T1-weighted images using a morphometric similarity network. White matter matrix was formed from diffusion tensor imaging-based structural connectivity. Multilayer network analysis of grey and white matter was performed using graph theory. We studied 29 isolated RBD patients and 30 healthy controls. Patients exhibited a higher average overlap degree (27.921 vs. 23.734, p = 0.002) and average multilayer clustering coefficient (0.474 vs. 0.413, p = 0.002) compared with controls. Additionally, several regions showed significant differences in the degree of overlap and multilayer clustering coefficient between patients with isolated RBD and healthy controls at the nodal level. The degree of overlap in the left medial orbitofrontal, left posterior cingulate, and right paracentral nodes and the multilayer clustering coefficients in the left lateral occipital, left rostral middle frontal, right fusiform, right inferior posterior parietal, and right parahippocampal nodes were higher in patients with isolated RBD than in healthy controls. We found alterations in the multilayer network at the global and nodal levels in patients with isolated RBD, and these changes may be associated with the pathophysiology of isolated RBD. Multilayer network analysis can be used widely to explore the mechanisms underlying various neurological disorders.

Community detection based on nonnegative matrix tri-factorization for multiplex social networks

Abstract Nowadays, community detection is one of the important fields for understanding network topology and has many applications in information diffusion, interaction mining and migration behaviour analysis. Therefore, community detection in social networks can help to understand user behaviour and network characteristics. There are many community detection methods, which are often designed for single-layer social networks. However, real-world networks use several types of relationships to establish connections between users, each of which has different characteristics. Hence, real-world networks can be modelled as multiplex networks. In general, multiplex networks are an example of multilayer networks in which the relationships between users in different networks can be considered simultaneously. In these networks, each layer represents the connections between users in a social network. Meanwhile, communities in multiplex networks are identified based on the structure and connections between overlapping users in different layers of the network. In this article, the nonnegative matrix tri-factorization (NMTF) strategy is used to model multiplex social networks, and a solution for community detection is developed based on it. According to this strategy, a common consensus matrix and then an alignment matrix are extracted based on similarity metrics and network structure. The use of these matrices as a flexible modelling framework enables the detection of coherent community between overlapping users in multiplex social networks. We evaluate the proposed NMTF method through various metrics on a multiplex social network. The results of this evaluation show the better performance of NMTF in terms of community quality compared to equivalent methods.

Classifying breast cancer using multi-view graph neural network based on multi-omics data.

Introduction: As the evaluation indices, cancer grading and subtyping have diverse clinical, pathological, and molecular characteristics with prognostic and therapeutic implications. Although researchers have begun to study cancer differentiation and subtype prediction, most of relevant methods are based on traditional machine learning and rely on single omics data. It is necessary to explore a deep learning algorithm that integrates multi-omics data to achieve classification prediction of cancer differentiation and subtypes. Methods: This paper proposes a multi-omics data fusion algorithm based on a multi-view graph neural network (MVGNN) for predicting cancer differentiation and subtype classification. The model framework consists of a graph convolutional network (GCN) module for learning features from different omics data and an attention module for integrating multi-omics data. Three different types of omics data are used. For each type of omics data, feature selection is performed using methods such as the chi-square test and minimum redundancy maximum relevance (mRMR). Weighted patient similarity networks are constructed based on the selected omics features, and GCN is trained using omics features and corresponding similarity networks. Finally, an attention module integrates different types of omics features and performs the final cancer classification prediction. Results: To validate the cancer classification predictive performance of the MVGNN model, we conducted experimental comparisons with traditional machine learning models and currently popular methods based on integrating multi-omics data using 5-fold cross-validation. Additionally, we performed comparative experiments on cancer differentiation and its subtypes based on single omics data, two omics data, and three omics data. Discussion: This paper proposed the MVGNN model and it performed well in cancer classification prediction based on multiple omics data.

Compound Similarity Network as a Novel Data Mining Strategy for High-Throughput Investigation of Degradation Pathways of Organic Pollutants in Industrial Wastewater Treatment.

Identification of degradation products and pathways is crucial for investigating emerging pollutants and evaluation of wastewater treatment methods. Nontargeted analysis is a powerful tool to comprehensively investigate the degradation pathways of organic pollutants in real-world wastewater samples but often generates large data sets, making it difficult to effectively locate the exact information on interests. Herein, to efficiently establish the linkages among compounds in the same degradation pathways, we introduce a compound similarity network (CSN) as a novel data mining strategy for LC-MS-based nontargeted analysis of complex wastewater samples. Different from molecular networks that cluster compounds based on MS/MS spectra similarity, our CSN strategy harnesses molecular fingerprints to establish linkages among compounds and thus is spectra-independent. The effectiveness of CSN was demonstrated by nontargeted identification of degradation pathways and products of organic pollutants in leather industrial wastewater that underwent laboratory-scale activated carbon adsorption (ACD) and ozonation treatments. Utilizing CSN in interpreting nontargeted data, we tentatively annotated 4324 compounds in the untreated leather industrial wastewater, 3246 after ACD, and 3777 after ACD/ozonation. We located 145 potential degradation pathways of organic pollutants in the ACD/ozonation process using CSN and validated 7 pathways with 15 chemical standards. CSN also revealed 5 clusters of emerging pollutants, from which 3 compounds were selected for in vitro cytotoxicity study to evaluate their potential biohazards as new pollutants. As CSN offers an efficient way to connect massive compounds and to find multiple degradation pathways in a high-throughput manner, we anticipate that it will find wide applications in nontargeted analysis of diverse environmental samples.

Prediction of MicroRNA-Disease Potential Association Based on Sparse Learning and Multilayer Random Walks.

More and more studies have shown that microRNAs (miRNAs) play an indispensable role in the study of complex diseases in humans. Traditional biological experiments to detect miRNA-disease associations are expensive and time-consuming. Therefore, it is necessary to propose efficient and meaningful computational models to predict miRNA-disease associations. In this study, we aim to propose a miRNA-disease association prediction model based on sparse learning and multilayer random walks (SLMRWMDA). The miRNA-disease association matrix is decomposed and reconstructed by the sparse learning method to obtain richer association information, and at the same time, the initial probability matrix for the random walk with restart algorithm is obtained. The disease similarity network, miRNA similarity network, and miRNA-disease association network are used to construct heterogeneous networks, and the stable probability is obtained based on the topological structure features of diseases and miRNAs through a multilayer random walk algorithm to predict miRNA-disease potential association. The experimental results show that the prediction accuracy of this model is significantly improved compared with the previous related models. We evaluated the model using global leave-one-out cross-validation (global LOOCV) and fivefold cross-validation (5-fold CV). The area under the curve (AUC) value for the LOOCV is 0.9368. The mean AUC value for 5-fold CV is 0.9335 and the variance is 0.0004. In the case study, the results show that SLMRWMDA is effective in inferring the potential association of miRNA-disease.

Ritual networks and the structure of moral communities in Classic Maya society

Ritual plays an important integrative function in the creation, maintenance, and transformation of human society. The shared experience of ritual establishes strong bonds between individuals that defines their membership in certain social groups. However, rituals are not timeless traditions, nor do they simply restore social equilibrium. Rather, rituals are active and ongoing social processes that unite and divide across multiple social categories. This paper applies archaeological network methods to analyze the multiscalar structure of ritual traditions across Classic Maya (ca. 300–900 CE) society using hieroglyphic inscriptions from dated and provenienced monuments cataloged in the Maya Hieroglyphic Database (Looper and Macri 1991–2024). For the Classic Maya, public ritual and performance were highly charged political events where meaning and power could be negotiated, creating opportunities for identity formation and community integration. Such contexts helped establish strong moral bonds in Classic Maya society. However, we know relatively little about the specific forms and content of these ritual practices. In this study we construct ritual similarity networks from hieroglyphic inscriptions to analyze the structure and organization of these moral communities as well as the ritual relations that bound them together.

PDDGCN: A Parasitic Disease-Drug Association Predictor Based on Multi-view Fusion Graph Convolutional Network.

The precise identification of associations between diseases and drugs is paramount for comprehending the etiology and mechanisms underlying parasitic diseases. Computational approaches are highly effective in discovering and predicting disease-drug associations. However, the majority of these approaches primarily rely on link-based methodologies within distinct biomedical bipartite networks. In this study, we reorganized a fundamental dataset of parasitic disease-drug associations using the latest databases, and proposed a prediction model called PDDGCN, based on a multi-view graph convolutional network. To begin with, we fused similarity networks with binary networks to establish multi-view heterogeneous networks. We utilized neighborhood information aggregation layers to refine node embeddings within each view of the multi-view heterogeneous networks, leveraging inter- and intra-domain message passing to aggregate information from neighboring nodes. Subsequently, we integrated multiple embeddings from each view and fed them into the ultimate discriminator. The experimental results demonstrate that PDDGCN outperforms five state-of-the-art methods and four compared machine learning algorithms. Additionally, case studies have substantiated the effectiveness of PDDGCN in identifying associations between parasitic diseases and drugs. In summary, the PDDGCN model has the potential to facilitate the discovery of potential treatments for parasitic diseases and advance our comprehension of the etiology in this field. The source code is available at https://github.com/AhauBioinformatics/PDDGCN .

A radiomics-based brain network in T1 images: construction, attributes, and applications.

T1 image is a widely collected imaging sequence in various neuroimaging datasets, but it is rarely used to construct an individual-level brain network. In this study, a novel individualized radiomics-based structural similarity network was proposed from T1 images. In detail, it used voxel-based morphometry to obtain the preprocessed gray matter images, and radiomic features were then extracted on each region of interest in Brainnetome atlas, and an individualized radiomics-based structural similarity network was finally built using the correlational values of radiomic features between any pair of regions of interest. After that, the network characteristics of individualized radiomics-based structural similarity network were assessed, including graph theory attributes, test-retest reliability, and individual identification ability (fingerprinting). At last, two representative applications for individualized radiomics-based structural similarity network, namely mild cognitive impairment subtype discrimination and fluid intelligence prediction, were exemplified and compared with some other networks on large open-source datasets. The results revealed that the individualized radiomics-based structural similarity network displays remarkable network characteristics and exhibits advantageous performances in mild cognitive impairment subtype discrimination and fluid intelligence prediction. In summary, the individualized radiomics-based structural similarity network provides a distinctive, reliable, and informative individualized structural brain network, which can be combined with other networks such as resting-state functional connectivity for various phenotypic and clinical applications.

Mononuclear phagocyte system-related multi-omics features yield head and neck squamous cell carcinoma subtypes with distinct overall survival, drug, and immunotherapy responses

BackgroundRecent research reported that mononuclear phagocyte system (MPS) can contribute to immune defense but the classification of head and neck squamous cell carcinoma (HNSCC) patients based on MPS-related multi-omics features using machine learning lacked.MethodsIn this study, we obtain marker genes for MPS through differential analysis at the single-cell level and utilize “similarity network fusion” and “MoCluster” algorithms to cluster patients’ multi-omics features. Subsequently, based on the corresponding clinical information, we investigate the prognosis, drugs, immunotherapy, and biological differences between the subtypes. A total of 848 patients have been included in this study, and the results obtained from the training set can be verified by two independent validation sets using “the nearest template prediction”.ResultsWe identified two subtypes of HNSCC based on MPS-related multi-omics features, with CS2 exhibiting better predictive prognosis and drug response. CS2 represented better xenobiotic metabolism and higher levels of T and B cell infiltration, while the biological functions of CS1 were mainly enriched in coagulation function, extracellular matrix, and the JAK-STAT signaling pathway. Furthermore, we established a novel and stable classifier called “getMPsub” to classify HNSCC patients, demonstrating good consistency in the same training set. External validation sets classified by “getMPsub” also illustrated similar differences between the two subtypes.ConclusionsOur study identified two HNSCC subtypes by machine learning and explored their biological difference. Notably, we constructed a robust classifier that presented an excellent classifying prediction, providing new insight into the precision medicine of HNSCC.

AlignScape, displaying sequence similarity using self-organizing maps.

The current richness of sequence data needs efficient methodologies to display and analyze the complexity of the information in a compact and readable manner. Traditionally, phylogenetic trees and sequence similarity networks have been used to display and analyze sequences of protein families. These methods aim to shed light on key computational biology problems such as sequence classification and functional inference. Here, we present a new methodology, AlignScape, based on self-organizing maps. AlignScape is applied to three large families of proteins: the kinases and GPCRs from human, and bacterial T6SS proteins. AlignScape provides a map of the similarity landscape and a tree representation of multiple sequence alignments These representations are useful to display, cluster, and classify sequences as well as identify functional trends. The efficient GPU implementation of AlignScape allows the analysis of large MSAs in a few minutes. Furthermore, we show how the AlignScape analysis of proteins belonging to the T6SS complex can be used to predict coevolving partners.

Improving enzyme functional annotation by integrating in vitro and in silico approaches: The example of histidinol phosphate phosphatases.

Advances in sequencing technologies have led to a rapid growth of public protein sequence databases, whereby the fraction of proteins with experimentally verified function continuously decreases. This problem is currently addressed by automated functional annotations with computational tools, which however lack the accuracy of experimental approaches and are susceptible to error propagation. Here, we present an approach that combines the efficiency of functional annotation by in silico methods with the rigor of enzyme characterization in vitro. First, a thorough experimental analysis of a representative enzyme of a group of homologues is performed which includes a focused alanine scan of the active site to determine a fingerprint of function-determining residues. In a second step, this fingerprint is used in combination with a sequence similarity network to identify putative isofunctional enzymes among the homologues. Using this approach in a proof-of-principle study, homologues of the histidinol phosphate phosphatase (HolPase) from Pseudomonas aeruginosa, many of which were annotated as phosphoserine phosphatases, were predicted to be HolPases. This functional annotation of the homologues was verified by in vitro testing of several representatives and an analysis of the occurrence of annotated HolPases in the corresponding phylogenetic groups. Moreover, the application of the same approach to the homologues of the HolPase from the archaeon Nitrosopumilus maritimus, which is not related to the HolPase from P. aeruginosa and was newly discovered in the course of this work, led to the annotation of the putative HolPase from various archaeal species.

ReHoGCNES-MDA: prediction of miRNA-disease associations using homogenous graph convolutional networks based on regular graph with random edge sampler.

Numerous investigations increasingly indicate the significance of microRNA (miRNA) in human diseases. Hence, unearthing associations between miRNA and diseases can contribute to precise diagnosis and efficacious remediation of medical conditions. The detection of miRNA-disease linkages via computational techniques utilizing biological information has emerged as a cost-effective and highly efficient approach. Here, we introduced a computational framework named ReHoGCNES, designed for prospective miRNA-disease association prediction (ReHoGCNES-MDA). This method constructs homogenous graph convolutional network with regular graph structure (ReHoGCN) encompassing disease similarity network, miRNA similarity network and known MDA network and then was tested on four experimental tasks. A random edge sampler strategy was utilized to expedite processes and diminish training complexity. Experimental results demonstrate that the proposed ReHoGCNES-MDA method outperforms both homogenous graph convolutional network and heterogeneous graph convolutional network with non-regular graph structure in all four tasks, which implicitly reveals steadily degree distribution of a graph does play an important role in enhancement of model performance. Besides, ReHoGCNES-MDA is superior to several machine learning algorithms and state-of-the-art methods on the MDA prediction. Furthermore, three case studies were conducted to further demonstrate the predictive ability of ReHoGCNES. Consequently, 93.3% (breast neoplasms), 90% (prostate neoplasms) and 93.3% (prostate neoplasms) of the top 30 forecasted miRNAs were validated by public databases. Hence, ReHoGCNES-MDA might serve as a dependable and beneficial model for predicting possible MDAs.

A semi-supervised approach for the integration of multi-omics data based on transformer multi-head self-attention mechanism and graph convolutional networks.

Comprehensive analysis of multi-omics data is crucial for accurately formulating effective treatment plans for complex diseases. Supervised ensemble methods have gained popularity in recent years for multi-omics data analysis. However, existing research based on supervised learning algorithms often fails to fully harness the information from unlabeled nodes and overlooks the latent features within and among different omics, as well as the various associations among features. Here, we present a novel multi-omics integrative method MOSEGCN, based on the Transformer multi-head self-attention mechanism and Graph Convolutional Networks(GCN), withtheaimofenhancingtheaccuracyofcomplexdiseaseclassification. MOSEGCN first employs the Transformer multi-head self-attention mechanism and Similarity Network Fusion (SNF) to separately learn the inherent correlations of latent features within and among different omics, constructing a comprehensive view of diseases. Subsequently, it feeds the learned crucial information into a self-ensembling Graph Convolutional Network (SEGCN) built upon semi-supervised learning methods for training and testing, facilitating a better analysis and utilization of information from multi-omics data to achieve precise classification of disease subtypes. The experimental results show that MOSEGCN outperforms several state-of-the-art multi-omics integrative analysis approaches on three types of omics data: mRNA expression data, microRNA expression data, and DNA methylation data, with accuracy rates of 83.0% for Alzheimer's disease and 86.7% for breast cancer subtyping. Furthermore, MOSEGCN exhibits strong generalizability on the GBM dataset, enabling the identification of important biomarkers for related diseases. MOSEGCN explores the significant relationship information among different omics and within each omics' latent features, effectively leveraging labeled and unlabeled information to further enhance the accuracy of complex disease classification. It also provides a promising approach for identifying reliable biomarkers, paving the way for personalized medicine.