The Comparative Toxicogenomics Database: update 2017

TIWMFLP: Two-Tier Interactive Weighted Matrix Factorization and Label Propagation Based on Similarity Matrix Fusion for Drug-Disease Association Prediction

Accurately identifying new therapeutic uses for drugs is crucial for advancing pharmaceutical research and development. Matrix factorization is often used in association prediction due to its simplicity and high interpretability. However, existing matrix factorization models do not enable real-time interaction between molecular feature matrices and similarity matrices, nor do they consider the geometric structure of the matrices. Additionally, efficiently integrating multisource data remains a significant challenge. To address these issues, we propose a two-tier interactive weighted matrix factorization and label propagation model based on similarity matrix fusion (TIWMFLP) to assist in personalized treatment. First, we calculate the Gaussian and Laplace kernel similarities for drugs and diseases using known drug-disease associations. We then introduce a new multisource similarity fusion method, called similarity matrix fusion (SMF), to integrate these drug/disease similarities. SMF not only considers the different contributions represented by each neighbor but also incorporates drug-disease association information to enhance the contextual topological relationships and potential features of each drug/disease node in the network. Second, we innovatively developed a two-tier interactive weighted matrix factorization (TIWMF) method to process three biological networks. This method realizes for the first time the real-time interaction between the drug/disease feature matrix and its similarity matrix, allowing for a better capture of the complex relationships between drugs and diseases. Additionally, the weighted matrix of the drug/disease similarity matrix is introduced to preserve the underlying structure of the similarity matrix. Finally, the label propagation algorithm makes predictions based on the three updated biological networks. Experimental outcomes reveal that TIWMFLP consistently surpasses state-of-the-art models on four drug-disease data sets, two small molecule-miRNA data sets, and one miRNA-disease data set.

Identification of senescence-related genes for potential therapeutic biomarkers of atrial fibrillation by bioinformatics, human histological validation, and molecular docking

Background

Cellular senescence is pivotal in the occurrence and progression of atrial fibrillation (AF). This study aimed to identify senescence-related genes that could be potential therapeutic biomarkers for AF.

Methods

AF-related differentially expressed genes (DEGs) were identified using the Gene Expression Omnibus dataset. Weighted gene co-expression network analysis (WGCNA) was used to analyze important modules and potential hub genes. Integrating senescence-related genes, potential biomarkers were identified. Their differential expression levels were then validated in human atrial tissue, HL-1 cells, and Angiotensin II-infused mice. Finally, molecular docking analysis was conducted to predict potential interactions between potential biomarkers and the senolytic drug Navitoclax.

Results

We identified seven genes common to AF-related DEGs and senescence-related genes. Three significant modules were selected from WGCNA analysis. Taken together, three senescence-related genes (ETS1, SP1, and WT1) were found to be significantly associated with AF. Protein-protein interaction network analysis revealed biological connections among the predicted target genes of ETS1, SP1, and WT1. Notably, ETS1, SP1, and WT1 exhibited significant differential expression in clinical samples as well as in vitro and in vivo models. Molecular docking revealed favorable binding affinity between senolytic Navitoclax and these potential biomarkers.

Conclusions

This study highlights ETS1, SP1, and WT1 as crucial senescence-related genes associated with AF, offering potential therapeutic targets, with supportive evidence of binding affinity with senolytic Navitoclax. These findings provide novel insights into AF pathogenesis from a senescence perspective.

Deciphering the spatial organization of fibrotic microenvironment in silica particles-induced pulmonary fibrosis

Silicosis represents a form of interstitial lung disease induced by the inhalation of silica particles in production environments. A key pathological characteristic of silica-induced pulmonary fibrosis is its localized tissue heterogeneity, which presents significant challenges in analyzing transcriptomic data due to the loss of important spatial context. To address this, we integrate spatial gene expression data with single-cell analyses and achieve a detailed mapping of cell types within and surrounding fibrotic regions, revealing significant shifts in cell populations in normal and diseased states. Additionally, we explore cell interactions within fibrotic zones using ligand-receptor mapping, deepening our understanding of cellular dynamics in these areas. We identify a subset of fibroblasts, termed Inmt fibroblasts, that play a suppressive role in the fibrotic microenvironment. Validating our findings through a comprehensive suite of bioinformatics, histological, and cell culture studies highlights the role of monocyte-derived macrophages in shifting Inmt fibroblast populations into profibrotic Grem1 fibroblast, potentially disrupting lung homeostasis in response to external challenges. Hence, the spatially detailed deconvolution offered by our research markedly advances the comprehension of cell dynamics and environmental interactions pivotal in the development of pulmonary fibrosis.

Identification of ENTPD1 as a novel biomarker linking allergic rhinitis and systemic lupus erythematosus

Several studies reveal that allergic rhinitis (AR) is a significant risk factor of systemic lupus erythematosus (SLE). However, studies investigating the common pathogenesis linking AR and SLE are lacking. Our study aims to search for the shared biomarkers and mechanisms that may provide new therapeutic targets for preventing AR from developing SLE. GSE50223 for AR and GSE103760 for SLE were downloaded from the Gene Expression Omnibus (GEO) database to screen differentially expressed genes (DEGs). The Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were performed to explore the functions of shared DEGs. Hub genes were screened by cytoHubba (a plugin of Cytoscape) and validated in another two datasets. Gene set enrichment analysis (GSEA) and single-sample Gene set enrichment analysis (ssGSEA) algorithm were applied to understand the functions of hub gene. ENTPD1 was validated as a hub gene between AR and SLE. GSEA results revealed that ENTPD1 was associated with KRAS_SIGNALING_UP pathway in AR and related to HYPOXIA, TGF_BETA_SIGNALING and TNFA_SIGNALING_VIA_NFKB pathways in SLE. The expression of ENTPD1 was positively correlated with activated CD8 T cell in both diseases. Thus, ENTPD1 may be a novel therapeutic target for preventing AR from developing SLE.

RACGAP1 is a pivotal gene in lung adenocarcinoma-associated membranous nephropathy: Based on comprehensive bioinformatics analysis and machine learning

BACKGROUND

This study performs a detailed bioinformatics and machine learning analysis to investigate the genetic foundations of membranous nephropathy (MN) in lung adenocarcinoma (LUAD).

METHODS

In this study, the gene expression profiles of MN microarray datasets (GSE99339) and LUAD dataset (GSE43767) were downloaded from the Gene Expression Omnibus database, common differentially expressed genes (DEGs) were obtained using the limma R package. The biological functions were analyzed with R Cluster Profiler package according to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. Machine learning algorithms, including LASSO regression, support vector machine (SVM), Random Forest, and Boruta analysis, were applied to identify hubgenes linked to LUAD-associated MN. These genes' prognostic values were evaluated in the TCGA-LUAD cohort and validated through immunohistochemistry on renal biopsy specimens.

RESULTS

A total of 36 DEGs in common were identified for downstream analyses. Functional enrichment analysis highlighted the involvement of the Toll-like receptor 4 pathway and several immune recognition pathways in LUAD-associated MN. COL3A1, PSENEN, RACGAP1, and TNFRSF10B were identified as hub genes in LUAD-associated MN using machine learning algorithms. ROC analysis demonstrated their effective discrimination of MN with high accuracy. Survival analysis showed that lung adenocarcinoma patients with higher expression of these genes had significantly reduced overall survival. In patients with lung adenocarcinoma-associated MN, RACGAP1, COL3A1, PSENEN, and TNFRSF10B were higher expressed in the glomerular, especially RACGAP1, indicating an important role in the pathogenesis of LUAD-associated membranous nephropathy.

CONCLUSIONS

Our study underscores the critical role of RACGAP1, COL3A1, PSENEN, and TNFRSF10B in the development of LUAD-associated MN, providing important insights for future research and the development of potential therapeutic strategies.

AI-driven Discovery of Morphomolecular Signatures in Toxicology

Early identification of drug toxicity is essential yet challenging in drug development. At the preclinical stage, toxicity is assessed with histopathological examination of tissue sections from animal models to detect morphological lesions. To complement this analysis, toxicogenomics is increasingly employed to understand the mechanism of action of the compound and ultimately identify lesion-specific safety biomarkers for which in vitro assays can be designed. However, existing works that aim to identify morphological correlates of expression changes rely on qualitative or semi-quantitative morphological characterization and remain limited in scale or morphological diversity. Artificial intelligence (AI) offers a promising approach for quantitatively modeling this relationship at an unprecedented scale. Here, we introduce GEESE, an AI model designed to impute morphomolecular signatures in toxicology data. Our model was trained to predict 1,536 gene targets on a cohort of 8,231 hematoxylin and eosin-stained liver sections from Rattus norvegicus across 127 preclinical toxicity studies. The model, evaluated on 2,002 tissue sections from 29 held-out studies, can yield pseudo-spatially resolved gene expression maps, which we correlate with six key drug-induced liver injuries (DILI). From the resulting 25 million lesion-expression pairs, we established quantitative relations between up and downregulated genes and lesions. Validation of these signatures against toxicogenomic databases, pathway enrichment analyses, and human hepatocyte cell lines asserted their relevance. Overall, our study introduces new methods for characterizing toxicity at an unprecedented scale and granularity, paving the way for AI-driven discovery of toxicity biomarkers.

HGBER: Heterogeneous Graph Neural Network With Bidirectional Encoding Representation

Heterogeneous graphs with multiple types of nodes and link relationships are ubiquitous in many real-world applications. Heterogeneous graph neural networks (HGNNs) as an efficient technique have shown superior capacity of dealing with heterogeneous graphs. Existing HGNNs usually define multiple meta-paths in a heterogeneous graph to capture the composite relations and guide neighbor selection. However, these models only consider the simple relationships (i.e., concatenation or linear superposition) between different meta-paths, ignoring more general or complex relationships. In this article, we propose a novel unsupervised framework termed Heterogeneous Graph neural network with bidirectional encoding representation (HGBER) to learn comprehensive node representations. Specifically, the contrastive forward encoding is firstly performed to extract node representations on a set of meta-specific graphs corresponding to meta-paths. We then introduce the reversed encoding for the degradation process from the final node representations to each single meta-specific node representations. Moreover, to learn structure-preserving node representations, we further utilize a self-training module to discover the optimal node distribution through iterative optimization. Extensive experiments on five open public datasets show that the proposed HGBER model outperforms the state-of-the-art HGNNs baselines by 0.8%-8.4% in terms of accuracy on most datasets in various downstream tasks.

Identification of Genes Crucial for Biological Processes in Breast Cancer Liver Metastasis Relapse

Breast cancer, when advancing to a metastatic stage, involves the liver, impacting over 50% of cases and significantly diminishing survival rates. Presently, a lack of tailored therapeutic protocols for breast cancer liver metastasis (BCLM) underscores the need for a deeper understanding of molecular patterns governing this complication. Therefore, by analyzing differentially expressed genes (DEGs) between primary breast tumors and BCLM lesions, we aimed to shed light on the diversities of this process. This research investigated breast cancer liver metastasis relapse by employing a comprehensive approach that integrated data filtering, gene ontology and KEGG pathway analysis, overall survival analysis, identification of the alteration in the DEGs, visualization of the protein-protein interaction network, Signor 2.0, identification of positively correlated genes, immune cell infiltration analysis, genetic alternation analysis, copy number variant analysis, gene-to-mRNA interaction, transcription factor analysis, molecular docking, and identification of potential treatment targets. This study's integrative approach unveiled metabolic reprogramming, suggesting altered PCK1 and LPL expression as key in breast cancer metastasis recurrence.

CAPN2 correlates with insulin resistance states in PCOS as evidenced by multi-dataset analysis

OBJECTIVE

IR emerges as a feature in the pathophysiology of PCOS, precipitating ovulatory anomalies and endometrial dysfunctions that contribute to the infertility challenges characteristic of this condition. Despite its clinical significance, a consensus on the precise mechanisms by which IR exacerbates PCOS is still lacking. This study aims to harness bioinformatics tools to unearth key IR-associated genes in PCOS patients, providing a platform for future therapeutic research and potential intervention strategies.

METHODS

We retrieved 4 datasets detailing PCOS from the GEO, and sourced IRGs from the MSigDB. We applied WGCNA to identify gene modules linked to insulin resistance, utilizing IR scores as a phenotypic marker. Gene refinement was executed through the LASSO, SVM, and Boruta feature selection algorithms. qPCR was carried out on selected samples to confirm findings. We predicted both miRNA and lncRNA targets using the ENCORI database, which facilitated the construction of a ceRNA network. Lastly, a drug-target network was derived from the CTD.

RESULTS

Thirteen genes related to insulin resistance in PCOS were identified via WGCNA analysis. LASSO, SVM, and Boruta algorithms further isolated CAPN2 as a notably upregulated gene, corroborated by biological verification. The ceRNA network involving lncRNA XIST and hsa-miR-433-3p indicated a possible regulatory link with CAPN2, supported by ENCORI database. Drug prediction analysis uncovered seven pharmacological agents, most being significant regulators of the endocrine system, as potential candidates for addressing insulin resistance in PCOS.

CONCLUSIONS

This study highlights the pivotal role of CAPN2 in insulin resistance within the context of PCOS, emphasizing its importance as both a critical biomarker and a potential therapeutic target. By identifying CAPN2, our research contributes to the expanding evidence surrounding the CAPN family, particularly CAPN10, in insulin resistance studies beyond PCOS. This work enriches our understanding of the mechanisms underlying insulin resistance, offering insights that bridge gaps in the current scientific landscape.

A probabilistic knowledge graph for target identification

Early identification of safe and efficacious disease targets is crucial to alleviating the tremendous cost of drug discovery projects. However, existing experimental methods for identifying new targets are generally labor-intensive and failure-prone. On the other hand, computational approaches, especially machine learning-based frameworks, have shown remarkable application potential in drug discovery. In this work, we propose Progeni, a novel machine learning-based framework for target identification. In addition to fully exploiting the known heterogeneous biological networks from various sources, Progeni integrates literature evidence about the relations between biological entities to construct a probabilistic knowledge graph. Graph neural networks are then employed in Progeni to learn the feature embeddings of biological entities to facilitate the identification of biologically relevant target candidates. A comprehensive evaluation of Progeni demonstrated its superior predictive power over the baseline methods on the target identification task. In addition, our extensive tests showed that Progeni exhibited high robustness to the negative effect of exposure bias, a common phenomenon in recommendation systems, and effectively identified new targets that can be strongly supported by the literature. Moreover, our wet lab experiments successfully validated the biological significance of the top target candidates predicted by Progeni for melanoma and colorectal cancer. All these results suggested that Progeni can identify biologically effective targets and thus provide a powerful and useful tool for advancing the drug discovery process.

A comparative benchmarking and evaluation framework for heterogeneous network-based drug repositioning methods

Computational drug repositioning, which involves identifying new indications for existing drugs, is an increasingly attractive research area due to its advantages in reducing both overall cost and development time. As a result, a growing number of computational drug repositioning methods have emerged. Heterogeneous network-based drug repositioning methods have been shown to outperform other approaches. However, there is a dearth of systematic evaluation studies of these methods, encompassing performance, scalability and usability, as well as a standardized process for evaluating new methods. Additionally, previous studies have only compared several methods, with conflicting results. In this context, we conducted a systematic benchmarking study of 28 heterogeneous network-based drug repositioning methods on 11 existing datasets. We developed a comprehensive framework to evaluate their performance, scalability and usability. Our study revealed that methods such as HGIMC, ITRPCA and BNNR exhibit the best overall performance, as they rely on matrix completion or factorization. HINGRL, MLMC, ITRPCA and HGIMC demonstrate the best performance, while NMFDR, GROBMC and SCPMF display superior scalability. For usability, HGIMC, DRHGCN and BNNR are the top performers. Building on these findings, we developed an online tool called HN-DREP (http://hn-drep.lyhbio.com/) to facilitate researchers in viewing all the detailed evaluation results and selecting the appropriate method. HN-DREP also provides an external drug repositioning prediction service for a specific disease or drug by integrating predictions from all methods. Furthermore, we have released a Snakemake workflow named HN-DRES (https://github.com/lyhbio/HN-DRES) to facilitate benchmarking and support the extension of new methods into the field.

Prevention of neointimal hyperplasia after coronary artery bypass graft via local delivery of sirolimus and rosuvastatin: network pharmacology and in vivo validation

BACKGROUND

Coronary artery bypass graft (CABG) is generally used to treat complex coronary artery disease. Treatment success is affected by neointimal hyperplasia (NIH) of graft and anastomotic sites. Although sirolimus and rosuvastatin individually inhibit NIH progression, the efficacy of combination treatment remains unknown.

METHODS

We identified cross-targets associated with CABG, sirolimus, and rosuvastatin by using databases including DisGeNET and GeneCards. GO and KEGG pathway enrichment analyses were conducted using R studio, and target proteins were mapped in PPI networks using Metascape and Cytoscape. For in vivo validation, we established a balloon-injured rabbit model by inducing NIH and applied a localized perivascular drug delivery device containing sirolimus and rosuvastatin. The outcomes were evaluated at 1, 2, and 4 weeks post-surgery.

RESULTS

We identified 115 shared targets between sirolimus and CABG among databases, 23 between rosuvastatin and CABG, and 96 among all three. TNF, AKT1, and MMP9 were identified as shared targets. Network pharmacology predicted the stages of NIH progression and the corresponding signaling pathways linked to sirolimus (acute stage, IL6/STAT3 signaling) and rosuvastatin (chronic stage, Akt/MMP9 signaling). In vivo experiments demonstrated that the combination of sirolimus and rosuvastatin significantly suppressed NIH progression. This combination treatment also markedly decreased the expression of inflammation and Akt signaling pathway-related proteins, which was consistent with the predictions from network pharmacology analysis.

CONCLUSIONS

Sirolimus and rosuvastatin inhibited pro-inflammatory cytokine production during the acute stage and regulated Akt/mTOR/NF-κB/STAT3 signaling in the chronic stage of NIH progression. These potential synergistic mechanisms may optimize treatment strategies to improve long-term patency after CABG.

Danggui Jixueteng decoction for the treatment of myelosuppression after chemotherapy: A combined metabolomics and network pharmacology analysis

Objective

This study aimed to explore the mechanism of the Danggui Jixueteng decoction (DJD) in treating Myelosuppression after chemotherapy (MAC) through network pharmacology and metabolomics.

Methods

We obtained the chemical structures of DJD compounds from TCMSP and PubMed. SwissTargetPrediction, STITCH, CTD, GeneCards, and OMIM were utilized to acquire component targets and MAC-related targets. We identified the key compounds, core targets, main biological processes, and signaling pathways related to DJD by constructing and analyzing related networks. The main active compounds and key proteins of DJD in treating AA were confirmed by molecular docking. A MAC rat model was established through intraperitoneal injection of cyclophosphamide to confirm DJD's effect on the bone marrow hematopoietic system. Untargeted metabolomics analyzed serum metabolite differences between MAC rats and the control group, and before and after DJD treatment, to explore DJD's mechanism in treating MAC.

Results

Of the 93 active compounds identified under screening conditions, 275 compound targets and 3113 MAC-related targets were obtained, including 95 intersecting targets; AKT1, STAT3, CASP3, and JUN were key proteins in MAC treatment. The phosphatidylinositol-3-kinase/RAC-alpha serine/threonine-protein kinase (PI3K/AKT) signaling pathway may play a crucial role in MAC treatment with DJD. Molecular docking results showed good docking effects of key protein AKT1 with luteolin, β-sitosterol, kaempferol, and glycyrrhizal chalcone A. In vivo experiments indicated that, compared to the model group, in the DJD group, levels of WBCs, RBCs, HGB, and PLTs in peripheral blood cells, thymus index increased, spleen index decreased, serum IL-3, GM-CSF levels increased, and IL-6, TNF-α, and VEGF levels decreased (p < 0.01); the pathological morphology of femoral bone marrow improved. Eleven differential metabolites were identified as differential serum metabolites, mainly concentrated in phenylalanine, tyrosine, and tryptophan biosynthesis pathways, phenylalanine metabolism, and arachidonic acid metabolism.

Conclusion

This study revealed that DJD's therapeutic effects are due to multiple ingredients, targets, and pathways. DJD may activate the PI3K/AKT signaling pathway, promote hematopoietic-related cytokine production, regulate related metabolic pathways, and effectively alleviate cyclophosphamide-induced myelosuppression after chemotherapy in rats.

Prediction of drug-disease associations based on reinforcement symmetric metric learning and graph convolution network

Accurately identifying novel indications for drugs is crucial in drug research and discovery. Traditional drug discovery is costly and time-consuming. Computational drug repositioning can provide an effective strategy for discovering potential drug-disease associations. However, the known experimentally verified drug-disease associations is relatively sparse, which may affect the prediction performance of the computational drug repositioning methods. Moreover, while the existing drug-disease prediction method based on metric learning algorithm has achieved better performance, it simply learns features of drugs and diseases only from the drug-centered perspective, and cannot comprehensively model the latent features of drugs and diseases. In this study, we propose a novel drug repositioning method named RSML-GCN, which applies graph convolutional network and reinforcement symmetric metric learning to predict potential drug-disease associations. RSML-GCN first constructs a drug-disease heterogeneous network by integrating the association and feature information of drugs and diseases. Then, the graph convolutional network (GCN) is applied to complement the drug-disease association information. Finally, reinforcement symmetric metric learning with adaptive margin is designed to learn the latent vector representation of drugs and diseases. Based on the learned latent vector representation, the novel drug-disease associations can be identified by the metric function. Comprehensive experiments on benchmark datasets demonstrated the superior prediction performance of RSML-GCN for drug repositioning.

Shared Proteins and Pathways of Cardiovascular and Cognitive Diseases: Relation to Vascular Cognitive Impairment

One of the primary goals of systems medicine is the detection of putative proteins and pathways involved in disease progression and pathological phenotypes. Vascular cognitive impairment (VCI) is a heterogeneous condition manifesting as cognitive impairment resulting from vascular factors. The precise mechanisms underlying this relationship remain unclear, which poses challenges for experimental research. Here, we applied computational approaches like systems biology to unveil and select relevant proteins and pathways related to VCI by studying the crosstalk between cardiovascular and cognitive diseases. In addition, we specifically included signals related to oxidative stress, a common etiologic factor tightly linked to aging, a major determinant of VCI. Our results show that pathways associated with oxidative stress are quite relevant, as most of the prioritized vascular cognitive genes and proteins were enriched in these pathways. Our analysis provided a short list of proteins that could be contributing to VCI: DOLK, TSC1, ATP1A1, MAPK14, YWHAZ, CREB3, HSPB1, PRDX6, and LMNA. Moreover, our experimental results suggest a high implication of glycative stress, generating oxidative processes and post-translational protein modifications through advanced glycation end-products (AGEs). We propose that these products interact with their specific receptors (RAGE) and Notch signaling to contribute to the etiology of VCI.

Bioinformatics Prediction for Network-Based Integrative Multi-Omics Expression Data Analysis in Hirschsprung Disease

Hirschsprung's disease (HSCR) is a rare developmental disorder in which enteric ganglia are missing along a portion of the intestine. HSCR has a complex inheritance, with RET as the major disease-causing gene. However, the pathogenesis of HSCR is still not completely understood. Therefore, we applied a computational approach based on multi-omics network characterization and clustering analysis for HSCR-related gene/miRNA identification and biomarker discovery. Protein-protein interaction (PPI) and miRNA-target interaction (MTI) networks were analyzed by DPClusO and BiClusO, respectively, and finally, the biomarker potential of miRNAs was computationally screened by miRNA-BD. In this study, a total of 55 significant gene-disease modules were identified, allowing us to propose 178 new HSCR candidate genes and two biological pathways. Moreover, we identified 12 key miRNAs with biomarker potential among 137 predicted HSCR-associated miRNAs. Functional analysis of new candidates showed that enrichment terms related to gene ontology (GO) and pathways were associated with HSCR. In conclusion, this approach has allowed us to decipher new clues of the etiopathogenesis of HSCR, although molecular experiments are further needed for clinical validations.

Comprehensive Review of Drug-Drug Interaction Prediction Based on Machine Learning: Current Status, Challenges, and Opportunities

Detecting drug-drug interactions (DDIs) is an essential step in drug development and drug administration. Given the shortcomings of current experimental methods, the machine learning (ML) approach has become a reliable alternative, attracting extensive attention from the academic and industrial fields. With the rapid development of computational science and the growing popularity of cross-disciplinary research, a large number of DDI prediction studies based on ML methods have been published in recent years. To give an insight into the current situation and future direction of DDI prediction research, we systemically review these studies from three aspects: (1) the classic DDI databases, mainly including databases of drugs, side effects, and DDI information; (2) commonly used drug attributes, which focus on chemical, biological, and phenotypic attributes for representing drugs; (3) popular ML approaches, such as shallow learning-based, deep learning-based, recommender system-based, and knowledge graph-based methods for DDI detection. For each section, related studies are described, summarized, and compared, respectively. In the end, we conclude the research status of DDI prediction based on ML methods and point out the existing issues, future challenges, potential opportunities, and subsequent research direction.

Drug repositioning with adaptive graph convolutional networks

MOTIVATION

Drug repositioning is an effective strategy to identify new indications for existing drugs, providing the quickest possible transition from bench to bedside. With the rapid development of deep learning, graph convolutional networks (GCNs) have been widely adopted for drug repositioning tasks. However, prior GCNs based methods exist limitations in deeply integrating node features and topological structures, which may hinder the capability of GCNs.

RESULTS

In this study, we propose an adaptive GCNs approach, termed AdaDR, for drug repositioning by deeply integrating node features and topological structures. Distinct from conventional graph convolution networks, AdaDR models interactive information between them with adaptive graph convolution operation, which enhances the expression of model. Concretely, AdaDR simultaneously extracts embeddings from node features and topological structures and then uses the attention mechanism to learn adaptive importance weights of the embeddings. Experimental results show that AdaDR achieves better performance than multiple baselines for drug repositioning. Moreover, in the case study, exploratory analyses are offered for finding novel drug-disease associations.

AVAILABILITY AND IMPLEMENTATION

The soure code of AdaDR is available at: https://github.com/xinliangSun/AdaDR.

Serum metabolic alterations in peritoneal dialysis patients with excessive daytime sleepiness

Excessive daytime sleepiness (EDS) is associated with quality of life and all-cause mortality in the end-stage renal disease population. This study aims to identify biomarkers and reveal the underlying mechanisms of EDS in peritoneal dialysis (PD) patients. A total of 48 nondiabetic continuous ambulatory peritoneal dialysis patients were assigned to the EDS group and the non-EDS group according to the Epworth Sleepiness Scale (ESS). Ultra-high-performance liquid chromatography coupled with quadrupole-time-of-flight mass spectrometry (UHPLC-Q-TOF/MS) was used to identify the differential metabolites. Twenty-seven (male/female, 15/12; age, 60.1 ± 16.2 years) PD patients with ESS ≥ 10 were assigned to the EDS group, while twenty-one (male/female, 13/8; age, 57.9 ± 10.1 years) PD patients with ESS < 10 were defined as the non-EDS group. With UHPLC-Q-TOF/MS, 39 metabolites with significant differences between the two groups were found, 9 of which had good correlations with disease severity and were further classified into amino acid, lipid and organic acid metabolism. A total of 103 overlapping target proteins of the differential metabolites and EDS were found. Then, the EDS-metabolite-target network and the protein-protein interaction network were constructed. The metabolomics approach integrated with network pharmacology provides new insights into the early diagnosis and mechanisms of EDS in PD patients.

Drug-disease association prediction using semantic graph and function similarity representation learning over heterogeneous information networks

Discovering new indications for existing drugs is a promising development strategy at various stages of drug research and development. However, most of them complete their tasks by constructing a variety of heterogeneous networks without considering available higher-order connectivity patterns in heterogeneous biological information networks, which are believed to be useful for improving the accuracy of new drug discovering. To this end, we propose a computational-based model, called SFRLDDA, for drug-disease association prediction by using semantic graph and function similarity representation learning. Specifically, SFRLDDA first integrates a heterogeneous information network (HIN) by drug-disease, drug-protein, protein-disease associations, and their biological knowledge. Second, different representation learning strategies are applied to obtain the feature representations of drugs and diseases from different perspectives over semantic graph and function similarity graphs constructed, respectively. At last, a Random Forest classifier is incorporated by SFRLDDA to discover potential drug-disease associations (DDAs). Experimental results demonstrate that SFRLDDA yields a best performance when compared with other state-of-the-art models on three benchmark datasets. Moreover, case studies also indicate that the simultaneous consideration of semantic graph and function similarity of drugs and diseases in the HIN allows SFRLDDA to precisely predict DDAs in a more comprehensive manner.

Systems toxicology of complex wood combustion aerosol reveals gaseous carbonyl compounds as critical constituents

Epidemiological studies identified air pollution as one of the prime causes for human morbidity and mortality, due to harmful effects mainly on the cardiovascular and respiratory systems. Damage to the lung leads to several severe diseases such as fibrosis, chronic obstructive pulmonary disease and cancer. Noxious environmental aerosols are comprised of a gas and particulate phase representing highly complex chemical mixtures composed of myriads of compounds. Although some critical pollutants, foremost particulate matter (PM), could be linked to adverse health effects, a comprehensive understanding of relevant biological mechanisms and detrimental aerosol constituents is still lacking. Here, we employed a systems toxicology approach focusing on wood combustion, an important source for air pollution, and demonstrate a key role of the gas phase, specifically carbonyls, in driving adverse effects. Transcriptional profiling and biochemical analysis of human lung cells exposed at the air-liquid-interface determined DNA damage and stress response, as well as perturbation of cellular metabolism, as major key events. Connectivity mapping revealed a high similarity of gene expression signatures induced by wood smoke and agents prompting DNA-protein crosslinks (DPCs). Indeed, various gaseous aldehydes were detected in wood smoke, which promote DPCs, initiate similar genomic responses and are responsible for DNA damage provoked by wood smoke. Hence, systems toxicology enables the discovery of critical constituents of complex mixtures i.e. aerosols and highlights the role of carbonyls on top of particulate matter as an important health hazard.

Predicting new drug indications based on double variational autoencoders

Experimental drug development is costly, complex, and time-consuming, and the number of drugs that have been put into application treatment is small. The identification of drug-disease correlations can provide important information for drug discovery and drug repurposing. Computational drug repurposing is an important and effective method that can be used to determine novel treatments for diseases. In recent years, an increasing number of large databases have been utilized for biological data research, particularly in the fields of drugs and diseases. Consequently, researchers have begun to explore the application of deep neural networks in biological data development. One particularly promising method for unsupervised learning is the deep generative model, with the variational autoencoder (VAE) being among the mainstream models. Here, we propose a drug indication prediction algorithm called DIDVAE (predicting new drug indications based on double variational autoencoders), which generates new data by learning the latent variable distribution of known data to achieve the goal of predicting drug-disease associations. In the experiment, we compared the DIDVAE algorithm with the BBNR, DrugNet, MBiRW and DRRS algorithms on a unified dataset. The comprehensive experimental results show that, compared with these prediction algorithms, the DIDVAE algorithm provides an overall improved prediction. In addition, further analysis and verification of the predicted unknown drug-disease association also proved the practicality of the method.

A Review on the Recent Applications of Deep Learning in Predictive Drug Toxicological Studies

Drug toxicity prediction is an important step in ensuring patient safety during drug design studies. While traditional preclinical studies have historically relied on animal models to evaluate toxicity, recent advances in deep-learning approaches have shown great promise in advancing drug safety science and reducing animal use in preclinical studies. However, deep-learning-based approaches also face challenges in handling large biological data sets, model interpretability, and regulatory acceptance. In this review, we provide an overview of recent developments in deep-learning-based approaches for predicting drug toxicity, highlighting their potential advantages over traditional methods and the need to address their limitations. Deep-learning models have demonstrated excellent performance in predicting toxicity outcomes from various data sources such as chemical structures, genomic data, and high-throughput screening assays. The potential of deep learning for automated feature engineering is also discussed. This review emphasizes the need to address ethical concerns related to the use of deep learning in drug toxicity studies, including the reduction of animal use and ensuring regulatory acceptance. Furthermore, emerging applications of deep learning in drug toxicity prediction, such as predicting drug-drug interactions and toxicity in rare subpopulations, are highlighted. The integration of deep-learning-based approaches with traditional methods is discussed as a way to develop more reliable and efficient predictive models for drug safety assessment, paving the way for safer and more effective drug discovery and development. Overall, this review highlights the critical role of deep learning in predictive toxicology and drug safety evaluation, emphasizing the need for continued research and development in this rapidly evolving field. By addressing the limitations of traditional methods, leveraging the potential of deep learning for automated feature engineering, and addressing ethical concerns, deep-learning-based approaches have the potential to revolutionize drug toxicity prediction and improve patient safety in drug discovery and development.

Drug vector representation and potential efficacy prediction based on graph representation learning and transcriptome data: Acacetin from traditional Chinese Medicine model

ETHNOPHARMACOLOGICAL RELEVANCE

Acacetin is widely distributed in traditional Chinese medicine and traditional herbs, with strong biological activity. Perhaps there are many potential effects that have not been explored. In the field of drug discovery, Mainstream methods focus on chemical structure. Traditional medicine cannot adapt to the mainstream prediction methods due to its complex composition.

AIM OF THE STUDY

Our aim is that provide a prediction method more suitable for traditional medicine by graph representation learning and transcriptome data. And use this method to predict acacetin.

MATERIALS AND METHODS

Our method mainly consists of two parts. The first part is to use the method of graph representation learning to vectorize drugs as a database. The original data of this part comes from transcriptome data on Gene Expression Omnibus. The method of graph representation learning is an unsupervised learning. If there is no prior knowledge as the label data, the training effect cannot be analyzed. Therefore, we define a standard score to evaluate our results through the idea of Jaccard index. The second part is to put the target drug into our database. The potential similarity between drugs was evaluated by the Euclidean distance between vectors, and the potential efficacy of the target drug is predicted by combining the chemical-disease relationship data in the Comparative Toxicogenomics Database. The target drug in this paper uses acacetin. We compared the predicted results with existing reports, and we also experimentally verified the efficacy of improving insulin resistance in the predicted results.

RESULTS

The prediction results are relatively consistent with the existing reports, which demonstrated that our method has a certain degree of predictive performance. And for the efficacy of improving insulin resistance in the predicted result, we verified it through experiments.

CONCLUSIONS

We propose a method to predict the potential efficacy of drugs based on transcriptome data, using Graph representation learning, which is very suitable for traditional medicine. Through this method, we predicted the efficacy of acacetin, and the results are relatively consistent with the current reports. This provides a new idea for unsupervised learning to apply medical information.

DTiGNN: Learning drug-target embedding from a heterogeneous biological network based on a two-level attention-based graph neural network

MOTIVATION

In vitro experiment-based drug-target interaction (DTI) exploration demands more human, financial and data resources. In silico approaches have been recommended for predicting DTIs to reduce time and cost. During the drug development process, one can analyze the therapeutic effect of the drug for a particular disease by identifying how the drug binds to the target for treating that disease. Hence, DTI plays a major role in drug discovery. Many computational methods have been developed for DTI prediction. However, the existing methods have limitations in terms of capturing the interactions via multiple semantics between drug and target nodes in a heterogeneous biological network (HBN).

METHODS

In this paper, we propose a DTiGNN framework for identifying unknown drug-target pairs. The DTiGNN first calculates the similarity between the drug and target from multiple perspectives. Then, the features of drugs and targets from each perspective are learned separately by using a novel method termed an information entropy-based random walk. Next, all of the learned features from different perspectives are integrated into a single drug and target similarity network by using a multi-view convolutional neural network. Using the integrated similarity networks, drug interactions, drug-disease associations, protein interactions and protein-disease association, the HBN is constructed. Next, a novel embedding algorithm called a meta-graph guided graph neural network is used to learn the embedding of drugs and targets. Then, a convolutional neural network is employed to infer new DTIs after balancing the sample using oversampling techniques.

RESULTS

The DTiGNN is applied to various datasets, and the result shows better performance in terms of the area under receiver operating characteristic curve (AUC) and area under precision-recall curve (AUPR), with scores of 0.98 and 0.99, respectively. There are 23,739 newly predicted DTI pairs in total.

The Environmental Conditions, Treatments, and Exposures Ontology (ECTO): connecting toxicology and exposure to human health and beyond

BACKGROUND

Evaluating the impact of environmental exposures on organism health is a key goal of modern biomedicine and is critically important in an age of greater pollution and chemicals in our environment. Environmental health utilizes many different research methods and generates a variety of data types. However, to date, no comprehensive database represents the full spectrum of environmental health data. Due to a lack of interoperability between databases, tools for integrating these resources are needed. In this manuscript we present the Environmental Conditions, Treatments, and Exposures Ontology (ECTO), a species-agnostic ontology focused on exposure events that occur as a result of natural and experimental processes, such as diet, work, or research activities. ECTO is intended for use in harmonizing environmental health data resources to support cross-study integration and inference for mechanism discovery.

METHODS AND FINDINGS

ECTO is an ontology designed for describing organismal exposures such as toxicological research, environmental variables, dietary features, and patient-reported data from surveys. ECTO utilizes the base model established within the Exposure Ontology (ExO). ECTO is developed using a combination of manual curation and Dead Simple OWL Design Patterns (DOSDP), and contains over 2700 environmental exposure terms, and incorporates chemical and environmental ontologies. ECTO is an Open Biological and Biomedical Ontology (OBO) Foundry ontology that is designed for interoperability, reuse, and axiomatization with other ontologies. ECTO terms have been utilized in axioms within the Mondo Disease Ontology to represent diseases caused or influenced by environmental factors, as well as for survey encoding for the Personalized Environment and Genes Study (PEGS).

CONCLUSIONS

We constructed ECTO to meet Open Biological and Biomedical Ontology (OBO) Foundry principles to increase translation opportunities between environmental health and other areas of biology. ECTO has a growing community of contributors consisting of toxicologists, public health epidemiologists, and health care providers to provide the necessary expertise for areas that have been identified previously as gaps.

Uncovering the genetic links of diabetic erectile dysfunction and chronic prostatitis/chronic pelvic pain syndrome

Background: Clinical associations between erectile dysfunction and chronic prostatitis/chronic pelvic pain syndrome (CP/CPPS) have been noticed, but the common pathogenic mechanisms between them remain elusive. The aim of the study was to mine shared genetic alterations between ED and chronic prostatitis/chronic pelvic pain syndrome. Method: Transcriptome data of ED and chronic prostatitis/chronic pelvic pain syndrome-related genes (CPRGs) were retrieved from relevant databases and differentially expressed analysis was used to obtain significant CPRGs. Then function enrichment and interaction analyses were performed to show shared transcriptional signature, including gene ontology and pathway enrichment, the construction of protein-protein interaction (PPI) network, cluster analysis, and co-expression analysis. Hub CPRGs and key cross-link were selected by validating these genes in clinical samples, chronic prostatitis/chronic pelvic pain syndrome and ED-related datasets. Then the miRNA-OSRGs co-regulatory network was predicted and validated. Subpopulation distribution and disease association of hub CPRGs were further identified. Result: Differentially expressed analysis revealed 363 significant CPRGs between ED and chronic prostatitis/chronic pelvic pain syndrome, functioning in inflammatory reaction, oxidative stress, apoptosis, smooth muscle cell proliferation, and extracellular matrix organization. A PPI network containing 245 nodes and 504 interactions was constructed. Module analysis depicted that multicellular organismal process and immune metabolic process were enriched. 17 genes were screened in PPI via topological algorithms, and reactive oxygen species as well as interleukin-1 metabolism were regarded as the bridging interactive mechanism. After screening and validation, a hub-CPRG signature consisting of COL1A1, MAPK6, LPL, NFE2L2 and NQO1 were identified and associated miRNA were verified. These miRNAs played an important role in immune and inflammatory response likewise. Finally, NQO1 was identified as a key genetic link between ED and chronic prostatitis/chronic pelvic pain syndrome. It was predominately enriched in corpus cavernosum endothelial cell, and correlated with other male urogenital and immune system diseases tightly. Conclusion: We identified the genetic profiles as well as corresponding regulatory network underlying interaction between ED and chronic prostatitis/chronic pelvic pain syndrome via multi-omics analysis. These findings expanded a new understanding for the molecular mechanism of ED with chronic prostatitis/chronic pelvic pain syndrome.

Tanshinone IIA promotes apoptosis by downregulating BCL2 and upregulating TP53 in triple-negative breast cancer

Tanshinone IIA (Tan IIA) was mainly used for cardiovascular disease treatment. Recent studies have demonstrated the role of Tan IIA for tumor treatment, but its mechanism remains unclear. At the first, the inhibitory effect of Tan IIA on 4T1 breast cancer cells was determined by CCK8 and colony formation assay. Then, a 4T1 BALB/c model of breast cancer was established to evaluate the anti-cancer effect of Tan IIA in vivo. Flow cytometry analysis and the TUNEL test were used to detect cell apoptosis in vitro and in vivo, respectively. The related targets and mechanisms of Tan IIA were predicted through network-based systems biology. At last, molecular docking and the molecular biological techniques were used to evaluate the predicted targets. Tan IIA displayed encouraging inhibitory influences on 4T1 cells after incubation for 24 h and showed a half-maximal inhibitory concentration (IC50) of 49.78 μM after 48-h incubation. After 23 days of treatment, the relative tumor volumes in the Tan IIA group were 65.53% inhibited compared with the control group. Furthermore, Tan IIA induced 4T1 cell apoptosis both in vivo and in vitro. The possible targets of Tan IIA for TNBC treatment were predicted with network-based systems biology, and results showed that TP53, NF-κB, AKT, MYC, and BCL-2 were the hub targets. The mechanism against breast cancer may be based on the P53 signaling pathway, the PI3K/Akt pathway, the MAPK signaling pathway, and the mTOR signaling pathways. Molecular docking analysis reveals that Tan IIA has a high affinity for p53, Bcl-2, and NF-κB1; the binding energies were - 6.92, - 6.07, and - 6.28 kcal/mol, respectively. The predicted proteins were further validated using Western blotting. Increased expression of phosphorylated p53 and p53 and decreased expression of Bcl-2 were found in Tan IIA-treated 4T1 cells. Tan IIA is potentially effective for the treatment of 4T1 breast cancer, and the molecular mechanism may be through enhancing the activity of p53 and decreasing Bcl-2 to suppress proliferation and promote apoptosis.

Synthesize heterogeneous biological knowledge via representation learning for Alzheimer's disease drug repurposing

Developing drugs for treating Alzheimer's disease has been extremely challenging and costly due to limited knowledge of underlying mechanisms and therapeutic targets. To address the challenge in AD drug development, we developed a multi-task deep learning pipeline that learns biological interactions and AD risk genes, then utilizes multi-level evidence on drug efficacy to identify repurposable drug candidates. Using the embedding derived from the model, we ranked drug candidates based on evidence from post-treatment transcriptomic patterns, efficacy in preclinical models, population-based treatment effects, and clinical trials. We mechanistically validated the top-ranked candidates in neuronal cells, identifying drug combinations with efficacy in reducing oxidative stress and safety in maintaining neuronal viability and morphology. Our neuronal response experiments confirmed several biologically efficacious drug combinations. This pipeline showed that harmonizing heterogeneous and complementary data/knowledge, including human interactome, transcriptome patterns, experimental efficacy, and real-world patient data shed light on the drug development of complex diseases.

Knowledge Guided Attention and Graph Convolutional Networks for Chemical-Disease Relation Extraction

The automatic extraction of the chemical-disease relation (CDR) from the text becomes critical because it takes a lot of time and effort to extract valuable CDR manually. Studies have shown that prior knowledge from the biomedical knowledge base is important for relation extraction. The method of combining deep learning models with prior knowledge is worthy of our study. In this paper, we propose a new model called Knowledge Guided Attention and Graph Convolutional Networks (KGAGN) for CDR extraction. First, to make full advantage of domain knowledge, we train entity embedding as a feature representation of input sequence, and relation embedding to capture weighted contextual information further through the attention mechanism. Then, to make full advantage of syntactic dependency information in cross-sentence CDR extraction, we construct document-level syntactic dependency graphs and encode them using a graph convolution network (GCN). Finally, the chemical-induced disease (CID) relation is extracted by using weighted context features and long-range dependency features both of which contain additional knowledge information We evaluated our model on the CDR dataset published by the BioCreative-V community and achieves an F1-score of 73.3%, surpassing other state-of-the-art methods. the code implemented by PyTorch 1.7.0 deep learning library can be downloaded from Github: https://github.com/sunyi123/cdr.

Exploration of the common genetic landscape of COVID-19 and male infertility

Background

COVID-19 has spread widely across continents since 2019, causing serious damage to human health. Accumulative research uncovered that SARS-CoV-2 poses a great threat to male fertility, and male infertility (MI) is a common comorbidity for the COVID-19 pandemic. The aim of the study was to explore the cross-talk molecular mechanisms between COVID-19 and MI.

Materials and methods

A total of four transcriptome data regarding COVID-19 and MI were downloaded from the Gene Expression Omnibus (GEO) repository, and were divided for two purposes (initial analysis and external validation). Differentially expressed genes (DEGs) analysis, GO and pathway annotation, protein-protein interaction (PPI) network, connectivity ranking, ROC analysis, immune infiltration, and translational and post-translational interaction were performed to gain hub COVID-19-related DEGs (CORGs). Moreover, we recorded medical information of COVID-19 patients with MI and matched healthy controls, and harvested their sperm samples in the university hospital. Expressions of hub CORGs were detected through the qRT-PCR technique.

Results

We identified 460 overlapped CORGs in both the COVID-19 DEGs and MI DEGs. CORGs were significantly enriched in DNA damage and repair-associated, cell cycle-associated, ubiquitination-associated, and coronavirus-associated signaling. Module assessment of PPI network revealed that enriched GO functions were closely related to cell cycle and DNA metabolism processes. Pharmacologic agent prediction displayed protein-drug interactions of ascorbic acid, biotin, caffeine, and L-cysteine with CORGs. After connectivity ranking and external validation, three hub CORGs (ENTPD6, CIB1, and EIF3B) showed good diagnostic performance (area under the curve > 0.75). Subsequently, three types of immune cells (CD8+ T cells, monocytes, and macrophages M0) were dominantly enriched, and 24 transcription factor-CORGs interactions and 13 miRNA-CORGs interactions were constructed in the network. Finally, qRT-PCR analysis confirmed that there were significant differences in the expression of hub CORGs (CIB1 and EIF3B) between the patient and control groups.

Conclusion

The present study identified and validated hub CORGs in COVID-19 and MI, and systematically explored molecular interactions and regulatory features in various biological processes. Our data provide new insights into the novel biomarkers and potential therapeutic targets of COVID-19-associated MI.

Identification of hub genes and biological mechanisms underlying the pathogenesis of asthenozoospermia and chronic epididymitis

Objective: Asthenozoospermia (AZS) is one of the most common causes of male fertility, affecting family wellbeing and population growth. Chronic epididymitis (CE) is a common and lingering inflammatory disease in the scrotum. Inflammation in the epididymis has a severe impact on sperm motility. This study aimed to explore the genetic profile and critical pathways involved in the pathological mechanisms of AZS and CE, and discover potential biomarkers. Methods: Genomic datasets of AZS and CE were obtained from the Gene Expression Omnibus (GEO) database, and relevant differentially expressed genes (DEGs) were identified. GO and pathway enrichment analyses, construction of a protein-protein interaction network, and receiver operator characteristic curve analysis were conducted. The expression profile of hub genes was validated in immunohistochemical data and testicular cell data. Immune infiltration, miRNA-hub gene interactions, and gene-disease interactions were explored. The mRNA levels of hub genes were further measured by qRT-PCR. Results: A total of 109 DEGs were identified between the AZS/CE and healthy control groups. Pathways of the immune system, neutrophil degranulation, and interleukin-4 and interleukin-13 signaling were enriched in AZS and CE. Five hub genes (CD300LB, CMKLR1, CCR4, B3GALT5, and CTSK) were selected, and their diagnostic values were validated in AZS, CE, and independent validation sets (area under the curve >0.7). Furthermore, the five-hub gene signature was well characterized in testicular immunohistochemical staining and testicular cells from healthy controls. Immune infiltration analysis showed that infiltration of CD8⁺ cells and T helper cells was significantly related to the expression level of five hub genes. In addition, a miRNA-hub gene network and interaction of other diseases were displayed. The mRNA levels of hub genes (CD300LB, CMKLR1, CCR4, and B3GALT5) were significantly elevated in the patient group. The mRNA level of CTSK also showed a similar trend. Conclusion: Our study uncovered the genetic profile involved in AZS and CE, and elucidated enriched pathways and molecular associations between hub genes and immune infiltration. This finding provides novel insight into the common pathogenesis of both diseases as well as the potential biomarkers for CE-associated AZS.

Prediction of potential small molecule-miRNA associations based on heterogeneous network representation learning

MicroRNAs (miRNAs) are closely associated with the occurrences and developments of many complex human diseases. Increasing studies have shown that miRNAs emerge as new therapeutic targets of small molecule (SM) drugs. Since traditional experiment methods are expensive and time consuming, it is particularly crucial to find efficient computational approaches to predict potential small molecule-miRNA (SM-miRNA) associations. Considering that integrating multi-source heterogeneous information related with SM-miRNA association prediction would provide a comprehensive insight into the features of both SMs and miRNAs, we proposed a novel model of Small Molecule-MiRNA Association prediction based on Heterogeneous Network Representation Learning (SMMA-HNRL) for more precisely predicting the potential SM-miRNA associations. In SMMA-HNRL, a novel heterogeneous information network was constructed with SM nodes, miRNA nodes and disease nodes. To access and utilize of the topological information of the heterogeneous information network, feature vectors of SM and miRNA nodes were obtained by two different heterogeneous network representation learning algorithms (HeGAN and HIN2Vec) respectively and merged with connect operation. Finally, LightGBM was chosen as the classifier of SMMA-HNRL for predicting potential SM-miRNA associations. The 10-fold cross validations were conducted to evaluate the prediction performance of SMMA-HNRL, it achieved an area under of ROC curve of 0.9875, which was superior to other three state-of-the-art models. With two independent validation datasets, the test experiment results revealed the robustness of our model. Moreover, three case studies were performed. As a result, 35, 37, and 22 miRNAs among the top 50 predicting miRNAs associated with 5-FU, cisplatin, and imatinib were validated by experimental literature works respectively, which confirmed the effectiveness of SMMA-HNRL. The source code and experimental data of SMMA-HNRL are available at https://github.com/SMMA-HNRL/SMMA-HNRL.

RLFDDA: a meta-path based graph representation learning model for drug-disease association prediction

BACKGROUND

Drug repositioning is a very important task that provides critical information for exploring the potential efficacy of drugs. Yet developing computational models that can effectively predict drug-disease associations (DDAs) is still a challenging task. Previous studies suggest that the accuracy of DDA prediction can be improved by integrating different types of biological features. But how to conduct an effective integration remains a challenging problem for accurately discovering new indications for approved drugs.

METHODS

In this paper, we propose a novel meta-path based graph representation learning model, namely RLFDDA, to predict potential DDAs on heterogeneous biological networks. RLFDDA first calculates drug-drug similarities and disease-disease similarities as the intrinsic biological features of drugs and diseases. A heterogeneous network is then constructed by integrating DDAs, disease-protein associations and drug-protein associations. With such a network, RLFDDA adopts a meta-path random walk model to learn the latent representations of drugs and diseases, which are concatenated to construct joint representations of drug-disease associations. As the last step, we employ the random forest classifier to predict potential DDAs with their joint representations.

RESULTS

To demonstrate the effectiveness of RLFDDA, we have conducted a series of experiments on two benchmark datasets by following a ten-fold cross-validation scheme. The results show that RLFDDA yields the best performance in terms of AUC and F1-score when compared with several state-of-the-art DDAs prediction models. We have also conducted a case study on two common diseases, i.e., paclitaxel and lung tumors, and found that 7 out of top-10 diseases and 8 out of top-10 drugs have already been validated for paclitaxel and lung tumors respectively with literature evidence. Hence, the promising performance of RLFDDA may provide a new perspective for novel DDAs discovery over heterogeneous networks.

Mechanisms of Cynarine for treatment of non-alcoholic fatty liver disease based on the integration of network pharmacology, molecular docking and cell experiment

BACKGROUND

Nonalcoholic Fatty Liver Disease (NAFLD) is a chronic Liver Disease prevalent all over the world. It has become more and more common in Japan, China and most western developed countries. The global prevalence rate is 25.24%, and the trend is increasing year by year. Related studies have shown that Cynarine has certain liver protection, lipid lowering and immune intervention effects. So, this study to systematically predict and analyze the mechanism of Cynarine in the treatment of non-alcoholic fatty liver disease (NAFLD) based on the integration of network pharmacology, molecular docking, and cell experiment.

METHODS

We performed Heatmap and Venn diagram analyses to identify genes and targets in Cynarine treat NAFLD. The network of Cynarine-therapeutic targets and the protein-protein interaction network (PPI) was constructed. We used gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses to visualize associated functional pathways. The Sybyl tool was used to dock the Cynarine with key therapeutic targets molecularly. Finally, cell experiments were applied to validate the role of Cynarine in the treatment of NAFLD.

RESULTS

The Cynarine could act on 48 targets of NAFLD, and the role of CASP3, TP53, MMP9, ELANE, NOTCH1 were more important. The PPI network showed that immune and inflammation-related targets played a pivotal role. The KEGG analysis found that the PI3K-Akt signaling pathway, cell cycle and MAPK signaling pathway may be the main pathways for Cynarine to prevent and treat NAFLD. Molecular docking studies confirmed that Cynarine has good binding activity with therapeutic targets. Cynarine reduced the fat deposition ability of NAFLD model cells, and effectively reduced the levels of ALT and AST released by liver cells due to excessive lipid accumulation. We also found that Cynarine inhibited the expression of AKT1 and MAPK1.

CONCLUSIONS

This study revealed that Cynarine could significantly reduce the fat deposition ability of NAFLD model cells, which may be closely related to the effective regulation of AKT1 and MAPK1 expression by Cynarine.

Druggable transcriptomic pathways revealed in Parkinson's patient-derived midbrain neurons

Complex genetic predispositions accelerate the chronic degeneration of midbrain substantia nigra neurons in Parkinson’s disease (PD). Deciphering the human molecular makeup of PD pathophysiology can guide the discovery of therapeutics to slow the disease progression. However, insights from human postmortem brain studies only portray the latter stages of PD, and there is a lack of data surrounding molecular events preceding the neuronal loss in patients. We address this gap by identifying the gene dysregulation of live midbrain neurons reprogrammed in vitro from the skin cells of 42 individuals, including sporadic and familial PD patients and matched healthy controls. To minimize bias resulting from neuronal reprogramming and RNA-seq methods, we developed an analysis pipeline integrating PD transcriptomes from different RNA-seq datasets (unsorted and sorted bulk vs. single-cell and Patch-seq) and reprogramming strategies (induced pluripotency vs. direct conversion). This PD cohort’s transcriptome is enriched for human genes associated with known clinical phenotypes of PD, regulation of locomotion, bradykinesia and rigidity. Dysregulated gene expression emerges strongest in pathways underlying synaptic transmission, metabolism, intracellular trafficking, neural morphogenesis and cellular stress/immune responses. We confirmed a synaptic impairment with patch-clamping and identified pesticides and endoplasmic reticulum stressors as the most significant gene-chemical interactions in PD. Subsequently, we associated the PD transcriptomic profile with candidate pharmaceuticals in a large database and a registry of current clinical trials. This study highlights human transcriptomic pathways that can be targeted therapeutically before the irreversible neuronal loss. Furthermore, it demonstrates the preclinical relevance of unbiased large transcriptomic assays of reprogrammed patient neurons.

Effect of Curcumin on Attenuation of Liver Cirrhosis via Genes/Proteins and Pathways: A System Pharmacology Study

Background: Liver cirrhosis is a life-threatening seqsuel of many chronic liver disorders of varying etiologies. In this study, we investigated protein targets of curcumin in liver cirrhosis based on a bioinformatics approach. Methods: Gene/protein associations with curcumin and liver cirrhosis were probed in drug−gene and gene−diseases databases including STITCH/DGIdb/DisGeNET/OMIM/DISEASES/CTD/Pharos and SwissTargetPrediction. Critical clustering groups (MCODE), hub candidates and critical hub genes in liver cirrhosis were identified, and connections between curcumin and liver cirrhosis-related genes were analyzed via Venn diagram. Interaction of hub genes with curcumin by molecular docking using PyRx-virtual screening tools was performed. Results: MCODE analysis indicated three MCODEs; the cluster (MCODE 1) comprised 79 nodes and 881 edges (score: 22.59). Curcumin database interactions recognized 318 protein targets. Liver cirrhosis genes and curcumin protein targets analysis demonstrated 96 shared proteins, suggesting that curcumin may influence 20 candidate and 13 hub genes, covering 81% of liver cirrhosis critical genes and proteins. Thirteen shared proteins affected oxidative stress regulation, RNA, telomerase activity, cell proliferation, and cell death. Molecular docking analysis showed the affinity of curcumin binding hub genes (Binding affinity: ΔG < −4.9 kcal/mol). Conclusions: Curcumin impacted on several critical liver cirrhosis genes mainly involved in extracellular matrix communication, focal adhesion, and the response to oxidative stress.

Heterogeneous Network Representation Learning: A Unified Framework with Survey and Benchmark

Since real-world objects and their interactions are often multi-modal and multi-typed, heterogeneous networks have been widely used as a more powerful, realistic, and generic superclass of traditional homogeneous networks (graphs). Meanwhile, representation learning (a.k.a. embedding) has recently been intensively studied and shown effective for various network mining and analytical tasks. In this work, we aim to provide a unified framework to deeply summarize and evaluate existing research on heterogeneous network embedding (HNE), which includes but goes beyond a normal survey. Since there has already been a broad body of HNE algorithms, as the first contribution of this work, we provide a generic paradigm for the systematic categorization and analysis over the merits of various existing HNE algorithms. Moreover, existing HNE algorithms, though mostly claimed generic, are often evaluated on different datasets. Understandable due to the application favor of HNE, such indirect comparisons largely hinder the proper attribution of improved task performance towards effective data preprocessing and novel technical design, especially considering the various ways possible to construct a heterogeneous network from real-world application data. Therefore, as the second contribution, we create four benchmark datasets with various properties regarding scale, structure, attribute/label availability, and etc. from different sources, towards handy and fair evaluations of HNE algorithms. As the third contribution, we carefully refactor and amend the implementations and create friendly interfaces for 13 popular HNE algorithms, and provide all-around comparisons among them over multiple tasks and experimental settings. By putting all existing HNE algorithms under a unified framework, we aim to provide a universal reference and guideline for the understanding and development of HNE algorithms. Meanwhile, by open-sourcing all data and code, we envision to serve the community with an ready-to-use benchmark platform to test and compare the performance of existing and future HNE algorithms (https://github.com/yangji9181/HNE).

Uncovering the pharmacology of Ginkgo biloba folium in the cell-type-specific targets of Parkinson's disease

Parkinson's disease (PD) is the second most common neurodegenerative disease with a fast-growing prevalence. Developing disease-modifying therapies for PD remains an enormous challenge. Current drug treatment will lose efficacy and bring about severe side effects as the disease progresses. Extracts from Ginkgo biloba folium (GBE) have been shown neuroprotective in PD models. However, the complex GBE extracts intertwingled with complicated PD targets hinder further drug development. In this study, we have pioneered using single-nuclei RNA sequencing data in network pharmacology analysis. Furthermore, high-throughput screening for potent drug-target interaction (DTI) was conducted with a deep learning algorithm, DeepPurpose. The strongest DTIs between ginkgolides and MAPK14 were further validated by molecular docking. This work should help advance the network pharmacology analysis procedure to tackle the limitation of conventional research. Meanwhile, these results should contribute to a better understanding of the complicated mechanisms of GBE in treating PD and lay the theoretical ground for future drug development in PD.

Preclinical insights into fucoidan as a nutraceutical compound against perfluorooctanoic acid-associated obesity via targeting endoplasmic reticulum stress

Obesity is a growing global health problem; it has been forecasted that over half of the global population will be obese by 2030. Obesity is complicated with many diseases, such as diabetes and cardiovascular diseases, leading to an economic impact on society. Other than diet, exposure to environmental pollutants is considered a risk factor for obesity. Exposure to perfluorooctanoic acid (PFOA) was found to impair hepatic lipid metabolism, resulting in obesity. In this study, we applied network pharmacology and systematic bioinformatics analysis, such as gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses, together with molecular docking, to investigate the targets of fucoidan for treating PFOA-associated obesity through the regulation of endoplasmic reticulum stress (ERS). Our results identified ten targets of fucoidan, such as glucosylceramidase beta (GBA), glutathione-disulfide reductase (GSR), melanocortin 4 receptor (MC4R), matrix metallopeptidase (MMP)2, MMP9, nuclear factor kappa B subunit 1 (NFKB1), RELA Proto-Oncogene, NF-KB Subunit (RELA), nuclear receptor subfamily 1 group I member 2 (NR1I2), proliferation-activated receptor delta (PPARD), and cellular retinoic acid binding protein 2 (CRABP2). GO and KEGG enrichment analyses highlighted their involvement in the pathogenesis of obesity, such as lipid and fat metabolisms. More importantly, the gene cluster is responsible for obesity-associated diseases and disorders, such as insulin resistance (IR), non-alcoholic fatty liver disease, and diabetic cardiomyopathy, via the control of signaling pathways. The findings of this report provide evidence that fucoidan is a potential nutraceutical product against PFOA-associated obesity through the regulation of ERS.

Identifying Potential Effective Diagnostic and Prognostic Biomarkers in Sepsis by Bioinformatics Analysis and Validation

Purpose

Sepsis is a serious life-threatening condition characterised by multi-organ failure due to a disturbed immune response caused by severe infection. The pathogenesis of sepsis is unclear. The aim of this article is to identify potential diagnostic and prognostic biomarkers of sepsis to improve the survival of patients with sepsis.

Methods

We downloaded 7 datasets from Gene Expression Omnibus database and screened the immune-related differential genes (IRDEGs). The related functions of IRDEGs were analyzed through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG). CIBERSORT was used to evaluate the infiltration of the immune cells, and Pearson algorithm of R software was used to calculate the correlation between the immune cell content and gene expression to screen the genes most related to immune cells in sepsis group, which were intersected with IRDEGs to obtain common genes. Key genes were identified from common genes based on the area under the receiver operating characteristic curve (AUC) greater than 0.8 in the 6 datasets. We then analyzed the predictive value of key genes in sepsis survival. Finally, we verified the expression of key genes in patients with sepsis by PCR analysis.

Results

A total of 164 IRDEGs were obtained, which were associated mainly with inflammatory and immunometabolic responses. Ten key genes (IL1R2, LTB4R, S100A11, S100A12, SORT1, RASGRP1, CD3G, CD40LG, CD8A and PPP3CC) were identified with high diagnostic efficacy. Logistic regression analysis revealed that six of the key genes (LTB4R, S100A11, SORT1, RASGRP1, CD3G and CD8A) may affect the survival prognosis of sepsis. PCR analysis confirmed that the expression of seven key genes (IL1R2, S100A12, RASGRP1, CD3G, CD40LG, CD8A and PPP3CC) was consistent with microarray outcome.

Conclusion

This study explored the immune and metabolic response mechanisms associated with sepsis, and identified ten potential diagnostic and six prognostic genes.

Identification of genes and key pathways underlying the pathophysiological association between nonalcoholic fatty liver disease and atrial fibrillation

Background

Atrial fibrillation (AF) is one of the most prevalent sustained cardiac arrhythmias. The latest studies have revealed a tight correlation between nonalcoholic fatty liver disease (NAFLD) and AF. However, the exact molecular mechanisms underlying the association between NAFLD and AF remain unclear. The current research aimed to expound the genes and signaling pathways that are related to the mechanisms underlying the association between these two diseases.

Materials and methods

NAFLD- and AF- related differentially expressed genes (DEGs) were identified via bioinformatic analysis of the Gene Expression Omnibus (GEO) datasets GSE63067 and GSE79768, respectively. Further enrichment analysis of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG), the construction of a protein–protein interaction (PPI) network, the identification of significant hub genes, and receiver operator characteristic curve analysis were conducted. The gene-disease interactions were analyzed using the Comparative Toxicogenomics Database. In addition, the hub genes were validated by quantitative Real-Time PCR (qRT-PCR) in NAFLD cell model.

Results

A total of 45 co-expressed differentially expressed genes (co-DEGs) were identified between the NAFLD/AF and healthy control individuals. GO and KEGG pathway analyses revealed that the co-DEGs were mostly enriched in neutrophil activation involved in the immune response and cytokine-cytokine receptor interactions. Moreover, eight hub genes were selected owing to their high degree of connectivity and upregulation in both the NAFLD and AF datasets. These genes included CCR2, PTPRC, CXCR2, MNDA, S100A9, NCF2, S100A12, and S100A8.

Conclusions

In summary, we conducted the gene differential expression analysis, functional enrichment analysis, and PPI analysis of DEGs in AF and NAFLD, which provides novel insights into the identification of potential biomarkers and valuable therapeutic leads for AF and NAFLD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12920-022-01300-1.

Identification of Colon Cancer-Related RNAs Based on Heterogeneous Networks and Random Walk

Colon cancer is considered as a complex disease that consists of metastatic seeding in early stages. Such disease is not simply caused by the action of a single RNA, but is associated with disorders of many kinds of RNAs and their regulation relationships. Hence, it is of great significance to study the complex regulatory roles among mRNAs, miRNAs and lncRNAs for further understanding the pathogenic mechanism of colon cancer. In this study, we constructed a heterogeneous network consisting of differentially expressed mRNAs, miRNAs and lncRNAs. This contains three kinds of vertices and six types of edges. All RNAs were re-divided into three categories, which were "related", "irrelevant" and "unlabeled". They were processed by dynamic excitation restart random walk (RW-DIR) for identifying colon cancer-related RNAs. Ten RNAs were finally obtained related to colon cancer, which were hsa-miR-2682-5p, hsa-miR-1277-3p, ANGPTL1, SLC22A18AS, FENDRR, PHLPP2, hsa-miR-302a-5p, APCDD1, MEX3A and hsa-miR-509-3-5p. Numerical experiments have indicated that the proposed network construction framework and the following RW-DIR algorithm are effective for identifying colon cancer-related RNAs, and this kind of analysis framework can also be easily extended to other diseases, effectively narrowing the scope of biological experimental research.

Role of statins in regulating molecular pathways following traumatic brain injury: A system pharmacology study

Traumatic brain injury (TBI) is a serious disorder with debilitating physical and psychological complications. Previous studies have indicated that genetic factors have a critical role in modulating the secondary phase of injury in TBI. Statins have interesting pleiotropic properties such as antiapoptotic, antioxidative, and anti-inflammatory effects, which make them a suitable class of drugs for repurposing in TBI. In this study, we aimed to explore how statins modulate proteins and pathways involved in TBI using system pharmacology. We first explored the target associations with statins in two databases to discover critical clustering groups, candidate hub and critical hub genes in the network of TBI, and the possible connections of statins with TBI-related genes. Our results showed 1763 genes associated with TBI. Subsequently, the analysis of centralities in the PPI network displayed 55 candidate hub genes and 15 hub genes. Besides, MCODE analysis based on threshold score:10 determined four modular clusters. Intersection analysis of genes related to TBI and statins demonstrated 204 shared proteins, which suggested that statins influence 31 candidate hub and 9 hub genes. Moreover, statins had the highest interaction with MCODE1. The biological processes of the 31 shared proteins are related to gene expression, inflammation, antioxidant activity, and cell proliferation. Biological enriched pathways showed Programmed Cell Death proteins, AGE-RAGE signaling pathway, C-type lectin receptor signalling pathway, and MAPK signaling pathway as top clusters. In conclusion, statins could target several critical post-TBI genes mainly involved in inflammation and apoptosis, supporting the previous research results as a potential therapeutic agent.

Mechanisms of Quercetin against atrial fibrillation explored by network pharmacology combined with molecular docking and experimental validation

Atrial fibrillation (AF) is a common atrial arrhythmia for which there is no specific therapeutic drug. Quercetin (Que) has been used to treat cardiovascular diseases such as arrhythmias. In this study, we explored the mechanism of action of Que in AF using network pharmacology and molecular docking. The chemical structure of Que was obtained from Pubchem. TCMSP, Swiss Target Prediction, Drugbank, STITCH, Pharmmapper, CTD, GeneCards, DISGENET and TTD were used to obtain drug component targets and AF-related genes, and extract AF and normal tissue by GEO database differentially expressed genes by GEO database. The top targets were IL6, VEGFA, JUN, MMP9 and EGFR, and Que for AF treatment might involve the role of AGE-RAGE signaling pathway in diabetic complications, MAPK signaling pathway and IL-17 signaling pathway. Molecular docking showed that Que binds strongly to key targets and is differentially expressed in AF. In vivo results showed that Que significantly reduced the duration of AF fibrillation and improved atrial remodeling, reduced p-MAPK protein expression, and inhibited the progression of AF. Combining network pharmacology and molecular docking approaches with in vivo studies advance our understanding of the intensive mechanisms of Quercetin, and provide the targeted basis for clinical Atrial fibrillation treatment.

Effects of exposure to oxytetracycline on the liver proteome of red seabream (Pagrus major) in a real administration scenario

Oxytetracycline (OTC) is a widely used antibiotic in aquaculture. In this study, red seabream (Pagrus major), the most popular aquaculture species in Japan, were treated with OTC mimicking a real administration scenario in aquaculture. The treatment groups were as follows: no OTC, 40 mg/kg body wt/day (equivalent to the dose used in actual aquaculture), or 178 mg/kg body wt/day. The first exposure was conducted for a week (1st OTC exposure period), followed by a 4-week interval, and the second exposure was for one week (2nd OTC exposure period). We investigated the effects of OTC on the liver proteome with the isobaric tags for relative and absolute quantitation (iTRAQ) technology accompanied by liquid chromatography and mass spectrometry. The pathway and disease enrichment analyses of differentially abundant proteins in OTC-exposed groups compared to their respective controls showed that the abundance of proteins related to the immune and nervous systems was altered after the 1st and 2nd OTC exposures, respectively. Quantitative real-time PCR of the transcripts of immune-related genes corroborated with the results of proteome analysis. OTC exposure also modulated the expression of metabolism-related proteins after the 1st and 2nd OTC exposures. Furthermore, after four weeks of the 2nd exposure, weight loss and changes in the expression of proteins related to metabolism were observed, suggesting that OTC exposure disrupts the metabolic system and causes growth inhibition. Based on these results, we suggest that the use of OTC in aquaculture poses a health risk in fish species. Thus, we need to pay more attention to the contamination with OTC in aquaculture.

Natural products for infectious microbes and diseases: an overview of sources, compounds, and chemical diversities

As coronavirus disease 2019 (COVID-19) threatens human health globally, infectious disorders have become one of the most challenging problem for the medical community. Natural products (NP) have been a prolific source of antimicrobial agents with widely divergent structures and a range vast biological activities. A dataset comprising 618 articles, including 646 NP-based compounds from 672 species of natural sources with biological activities against 21 infectious pathogens from five categories, was assembled through manual selection of published articles. These data were used to identify 268 NP-based compounds classified into ten groups, which were used for network pharmacology analysis to capture the most promising lead-compounds such as agelasine D, dicumarol, dihydroartemisinin and pyridomycin. The distribution of maximum Tanimoto scores indicated that compounds which inhibited parasites exhibited low diversity, whereas the chemistries inhibiting bacteria, fungi, and viruses showed more structural diversity. A total of 331 species of medicinal plants with compounds exhibiting antimicrobial activities were selected to classify the family sources. The family Asteraceae possesses various compounds against C. neoformans, the family Anacardiaceae has compounds against Salmonella typhi, the family Cucurbitacea against the human immunodeficiency virus (HIV), and the family Ancistrocladaceae against Plasmodium. This review summarizes currently available data on NP-based antimicrobials against refractory infections to provide information for further discovery of drugs and synthetic strategies for anti-infectious agents.

Applications of knowledge graphs for food science and industry

The deployment of various networks (e.g., Internet of Things [IoT] and mobile networks), databases (e.g., nutrition tables and food compositional databases), and social media (e.g., Instagram and Twitter) generates huge amounts of food data, which present researchers with an unprecedented opportunity to study various problems and applications in food science and industry via data-driven computational methods. However, these multi-source heterogeneous food data appear as information silos, leading to difficulty in fully exploiting these food data. The knowledge graph provides a unified and standardized conceptual terminology in a structured form, and thus can effectively organize these food data to benefit various applications. In this review, we provide a brief introduction to knowledge graphs and the evolution of food knowledge organization mainly from food ontology to food knowledge graphs. We then summarize seven representative applications of food knowledge graphs, such as new recipe development, diet-disease correlation discovery, and personalized dietary recommendation. We also discuss future directions in this field, such as multimodal food knowledge graph construction and food knowledge graphs for human health.

Explore potential disease related metabolites based on latent factor model

Background

In biological systems, metabolomics can not only contribute to the discovery of metabolic signatures for disease diagnosis, but is very helpful to illustrate the underlying molecular disease-causing mechanism. Therefore, identification of disease-related metabolites is of great significance for comprehensively understanding the pathogenesis of diseases and improving clinical medicine.

Results

In the paper, we propose a disease and literature driven metabolism prediction model (DLMPM) to identify the potential associations between metabolites and diseases based on latent factor model. We build the disease glossary with disease terms from different databases and an association matrix based on the mapping between diseases and metabolites. The similarity of diseases and metabolites is used to complete the association matrix. Finally, we predict potential associations between metabolites and diseases based on the matrix decomposition method. In total, 1,406 direct associations between diseases and metabolites are found. There are 119,206 unknown associations between diseases and metabolites predicted with a coverage rate of 80.88%. Subsequently, we extract training sets and testing sets based on data increment from the database of disease-related metabolites and assess the performance of DLMPM on 19 diseases. As a result, DLMPM is proven to be successful in predicting potential metabolic signatures for human diseases with an average AUC value of 82.33%.

Conclusion

In this paper, a computational model is proposed for exploring metabolite-disease pairs and has good performance in predicting potential metabolites related to diseases through adequate validation. The results show that DLMPM has a better performance in prioritizing candidate diseases-related metabolites compared with the previous methods and would be helpful for researchers to reveal more information about human diseases.