Functional diversity of topological modules in human protein-protein interaction networks

KDGene: knowledge graph completion for disease gene prediction using interactional tensor decomposition

The accurate identification of disease-associated genes is crucial for understanding the molecular mechanisms underlying various diseases. Most current methods focus on constructing biological networks and utilizing machine learning, particularly deep learning, to identify disease genes. However, these methods overlook complex relations among entities in biological knowledge graphs. Such information has been successfully applied in other areas of life science research, demonstrating their effectiveness. Knowledge graph embedding methods can learn the semantic information of different relations within the knowledge graphs. Nonetheless, the performance of existing representation learning techniques, when applied to domain-specific biological data, remains suboptimal. To solve these problems, we construct a biological knowledge graph centered on diseases and genes, and develop an end-to-end knowledge graph completion framework for disease gene prediction using interactional tensor decomposition named KDGene. KDGene incorporates an interaction module that bridges entity and relation embeddings within tensor decomposition, aiming to improve the representation of semantically similar concepts in specific domains and enhance the ability to accurately predict disease genes. Experimental results show that KDGene significantly outperforms state-of-the-art algorithms, whether existing disease gene prediction methods or knowledge graph embedding methods for general domains. Moreover, the comprehensive biological analysis of the predicted results further validates KDGene's capability to accurately identify new candidate genes. This work proposes a scalable knowledge graph completion framework to identify disease candidate genes, from which the results are promising to provide valuable references for further wet experiments. Data and source codes are available at https://github.com/2020MEAI/KDGene.

Protein interaction networks provide insight into fetal origins of chronic obstructive pulmonary disease

BACKGROUND

Chronic obstructive pulmonary disease (COPD) is a leading cause of death in adults that may have origins in early lung development. It is a complex disease, influenced by multiple factors including genetic variants and environmental factors. Maternal smoking during pregnancy may influence the risk for diseases during adulthood, potentially through epigenetic modifications including methylation.

METHODS

In this work, we explore the fetal origins of COPD by utilizing lung DNA methylation marks associated with in utero smoke (IUS) exposure, and evaluate the network relationships between methylomic and transcriptomic signatures associated with adult lung tissue from former smokers with and without COPD. To identify potential pathobiological mechanisms that may link fetal lung, smoke exposure and adult lung disease, we study the interactions (physical and functional) of identified genes using protein-protein interaction networks.

RESULTS

We build IUS-exposure and COPD modules, which identify connected subnetworks linking fetal lung smoke exposure to adult COPD. Studying the relationships and connectivity among the different modules for fetal smoke exposure and adult COPD, we identify enriched pathways, including the AGE-RAGE and focal adhesion pathways.

CONCLUSIONS

The modules identified in our analysis add new and potentially important insights to understanding the early life molecular perturbations related to the pathogenesis of COPD. We identify AGE-RAGE and focal adhesion as two biologically plausible pathways that may reveal lung developmental contributions to COPD. We were not only able to identify meaningful modules but were also able to study interconnections between smoke exposure and lung disease, augmenting our knowledge about the fetal origins of COPD.

PDGNet: Predicting Disease Genes Using a Deep Neural Network With Multi-View Features

The knowledge of phenotype-genotype associations is crucial for the understanding of disease mechanisms. Numerous studies have focused on developing efficient and accurate computing approaches to predict disease genes. However, owing to the sparseness and complexity of medical data, developing an efficient deep neural network model to identify disease genes remains a huge challenge. Therefore, we develop a novel deep neural network model that fuses the multi-view features of phenotypes and genotypes to identify disease genes (termed PDGNet). Our model integrated the multi-view features of diseases and genes and leveraged the feedback information of training samples to optimize the parameters of deep neural network and obtain the deep vector features of diseases and genes. The evaluation experiments on a large data set indicated that PDGNet obtained higher performance than the state-of-the-art method (precision and recall improved by 9.55 and 9.63 percent). The analysis results for the candidate genes indicated that the predicted genes have strong functional homogeneity and dense interactions with known genes. We validated the top predicted genes of Parkinson's disease based on external curated data and published medical literatures, which indicated that the candidate genes have a huge potential to guide the selection of causal genes in the 'wet experiment'. The source codes and the data of PDGNet are available at https://github.com/yangkuoone/PDGNet.

Performance Assessment of the Network Reconstruction Approaches on Various Interactomes

Beyond the list of molecules, there is a necessity to collectively consider multiple sets of omic data and to reconstruct the connections between the molecules. Especially, pathway reconstruction is crucial to understanding disease biology because abnormal cellular signaling may be pathological. The main challenge is how to integrate the data together in an accurate way. In this study, we aim to comparatively analyze the performance of a set of network reconstruction algorithms on multiple reference interactomes. We first explored several human protein interactomes, including PathwayCommons, OmniPath, HIPPIE, iRefWeb, STRING, and ConsensusPathDB. The comparison is based on the coverage of each interactome in terms of cancer driver proteins, structural information of protein interactions, and the bias toward well-studied proteins. We next used these interactomes to evaluate the performance of network reconstruction algorithms including all-pair shortest path, heat diffusion with flux, personalized PageRank with flux, and prize-collecting Steiner forest (PCSF) approaches. Each approach has its own merits and weaknesses. Among them, PCSF had the most balanced performance in terms of precision and recall scores when 28 pathways from NetPath were reconstructed using the listed algorithms. Additionally, the reference interactome affects the performance of the network reconstruction approaches. The coverage and disease- or tissue-specificity of each interactome may vary, which may result in differences in the reconstructed networks.

Entropy-Based Graph Clustering of PPI Networks for Predicting Overlapping Functional Modules of Proteins

Functional modules can be predicted using genome-wide protein-protein interactions (PPIs) from a systematic perspective. Various graph clustering algorithms have been applied to PPI networks for this task. In particular, the detection of overlapping clusters is necessary because a protein is involved in multiple functions under different conditions. graph entropy (GE) is a novel metric to assess the quality of clusters in a large, complex network. In this study, the unweighted and weighted GE algorithm is evaluated to prove the validity of predicting function modules. To measure clustering accuracy, the clustering results are compared to protein complexes and Gene Ontology (GO) annotations as references. We demonstrate that the GE algorithm is more accurate in overlapping clusters than the other competitive methods. Moreover, we confirm the biological feasibility of the proteins that occur most frequently in the set of identified clusters. Finally, novel proteins for the additional annotation of GO terms are revealed.

Knowledge-Guided "Community Network" Analysis Reveals the Functional Modules and Candidate Targets in Non-Small-Cell Lung Cancer

Non-small-cell lung cancer (NSCLC) represents a heterogeneous group of malignancies that are the leading cause of cancer-related death worldwide. Although many NSCLC-related genes and pathways have been identified, there remains an urgent need to mechanistically understand how these genes and pathways drive NSCLC. Here, we propose a knowledge-guided and network-based integration method, called the node and edge Prioritization-based Community Analysis, to identify functional modules and their candidate targets in NSCLC. The protein–protein interaction network was prioritized by performing a random walk with restart algorithm based on NSCLC seed genes and the integrating edge weights, and then a “community network” was constructed by combining Girvan–Newman and Label Propagation algorithms. This systems biology analysis revealed that the CCNB1-mediated network in the largest community provides a modular biomarker, the second community serves as a drug regulatory module, and the two are connected by some contextual signaling motifs. Moreover, integrating structural information into the signaling network suggested novel protein–protein interactions with therapeutic significance, such as interactions between GNG11 and CXCR2, CXCL3, and PPBP. This study provides new mechanistic insights into the landscape of cellular functions in the context of modular networks and will help in developing therapeutic targets for NSCLC.

A network-based machine-learning framework to identify both functional modules and disease genes

Disease gene identification is a critical step towards uncovering the molecular mechanisms of diseases and systematically investigating complex disease phenotypes. Despite considerable efforts to develop powerful computing methods, candidate gene identification remains a severe challenge owing to the connectivity of an incomplete interactome network, which hampers the discovery of true novel candidate genes. We developed a network-based machine-learning framework to identify both functional modules and disease candidate genes. In this framework, we designed a semi-supervised non-negative matrix factorization model to obtain the functional modules related to the diseases and genes. Of note, we proposed a disease gene-prioritizing method called MapGene that integrates the correlations from both functional modules and network closeness. Our framework identified a set of functional modules with highly functional homogeneity and close gene interactions. Experiments on a large-scale benchmark dataset showed that MapGene performs significantly better than the state-of-the-art algorithms. Further analysis demonstrates MapGene can effectively relieve the impact of the incompleteness of interactome networks and obtain highly reliable rankings of candidate genes. In addition, disease cases on Parkinson's disease and diabetes mellitus confirmed the generalization of MapGene for novel candidate gene identification. This work proposed, for the first time, an integrated computing framework to predict both functional modules and disease candidate genes. The methodology and results support that our framework has the potential to help discover underlying functional modules and reliable candidate genes in human disease.

It is time to apply biclustering: a comprehensive review of biclustering applications in biological and biomedical data

Biclustering is a powerful data mining technique that allows clustering of rows and columns, simultaneously, in a matrix-format data set. It was first applied to gene expression data in 2000, aiming to identify co-expressed genes under a subset of all the conditions/samples. During the past 17 years, tens of biclustering algorithms and tools have been developed to enhance the ability to make sense out of large data sets generated in the wake of high-throughput omics technologies. These algorithms and tools have been applied to a wide variety of data types, including but not limited to, genomes, transcriptomes, exomes, epigenomes, phenomes and pharmacogenomes. However, there is still a considerable gap between biclustering methodology development and comprehensive data interpretation, mainly because of the lack of knowledge for the selection of appropriate biclustering tools and further supporting computational techniques in specific studies. Here, we first deliver a brief introduction to the existing biclustering algorithms and tools in public domain, and then systematically summarize the basic applications of biclustering for biological data and more advanced applications of biclustering for biomedical data. This review will assist researchers to effectively analyze their big data and generate valuable biological knowledge and novel insights with higher efficiency.

Network-based method for drug target discovery at the isoform level

Identification of primary targets associated with phenotypes can facilitate exploration of the underlying molecular mechanisms of compounds and optimization of the structures of promising drugs. However, the literature reports limited effort to identify the target major isoform of a single known target gene. The majority of genes generate multiple transcripts that are translated into proteins that may carry out distinct and even opposing biological functions through alternative splicing. In addition, isoform expression is dynamic and varies depending on the developmental stage and cell type. To identify target major isoforms, we integrated a breast cancer type-specific isoform coexpression network with gene perturbation signatures in the MCF7 cell line in the Connectivity Map database using the ‘shortest path’ drug target prioritization method. We used a leukemia cancer network and differential expression data for drugs in the HL-60 cell line to test the robustness of the detection algorithm for target major isoforms. We further analyzed the properties of target major isoforms for each multi-isoform gene using pharmacogenomic datasets, proteomic data and the principal isoforms defined by the APPRIS and STRING datasets. Then, we tested our predictions for the most promising target major protein isoforms of DNMT1, MGEA5 and P4HB4 based on expression data and topological features in the coexpression network. Interestingly, these isoforms are not annotated as principal isoforms in APPRIS. Lastly, we tested the affinity of the target major isoform of MGEA5 for streptozocin through in silico docking. Our findings will pave the way for more effective and targeted therapies via studies of drug targets at the isoform level.

Discovery and characterization of functional modules associated with body weight in broilers

Aim of the present study was to investigate whether body weight (BW) in broilers is associated with functional modular genes. To this end, first a GWAS for BW was conducted using 6,598 broilers and the high density SNP array. The next step was to search for positional candidate genes and QTLs within strong LD genomic regions around the significant SNPs. Using all positional candidate genes, a network was then constructed and community structure analysis was performed. Finally, functional enrichment analysis was applied to infer the functional relevance of modular genes. A total number of 645 positional candidate genes were identified in strong LD genomic regions around 11 genome-wide significant markers. 428 of the positional candidate genes were located within growth related QTLs. Community structure analysis detected 5 modules while functional enrichment analysis showed that 52 modular genes participated in developmental processes such as skeletal system development. An additional number of 14 modular genes (GABRG1, NGF, APOBEC2, STAT5B, STAT3, SMAD4, MED1, CACNB1, SLAIN2, LEMD2, ZC3H18, TMEM132D, FRYL and SGCB) were also identified as related to body weight. Taken together, current results suggested a total number of 66 genes as most plausible functional candidates for the trait examined.

A Comparative Study of Cluster Detection Algorithms in Protein-Protein Interaction for Drug Target Discovery and Drug Repurposing

The interactions between drugs and their target proteins induce altered expression of genes involved in complex intracellular networks. The properties of these functional network modules are critical for the identification of drug targets, for drug repurposing, and for understanding the underlying mode of action of the drug. The topological modules generated by a computational approach are defined as functional clusters. However, the functions inferred for these topological modules extracted from a large-scale molecular interaction network, such as a protein–protein interaction (PPI) network, could differ depending on different cluster detection algorithms. Moreover, the dynamic gene expression profiles among tissues or cell types causes differential functional interaction patterns between the molecular components. Thus, the connections in the PPI network should be modified by the transcriptomic landscape of specific cell lines before producing topological clusters. Here, we systematically investigated the clusters of a cell-based PPI network by using four cluster detection algorithms. We subsequently compared the performance of these algorithms for target gene prediction, which integrates gene perturbation data with the cell-based PPI network using two drug target prioritization methods, shortest path and diffusion correlation. In addition, we validated the proportion of perturbed genes in clusters by finding candidate anti-breast cancer drugs and confirming our predictions using literature evidence and cases in the ClinicalTrials.gov. Our results indicate that the Walktrap (CW) clustering algorithm achieved the best performance overall in our comparative study.

Functional homogeneity and specificity of topological modules in human proteome

BACKGROUND

Functional modules in protein-protein interaction networks (PPIN) are defined by maximal sets of functionally associated proteins and are vital to understanding cellular mechanisms and identifying disease associated proteins. Topological modules of the human proteome have been shown to be related to functional modules of PPIN. However, the effects of the weights of interactions between protein pairs and the integration of physical (direct) interactions with functional (indirect expression-based) interactions have not been investigated in the detection of functional modules of the human proteome.

RESULTS

We investigated functional homogeneity and specificity of topological modules of the human proteome and validated them with known biological and disease pathways. Specifically, we determined the effects on functional homogeneity and heterogeneity of topological modules (i) with both physical and functional protein-protein interactions; and (ii) with incorporation of functional similarities between proteins as weights of interactions. With functional enrichment analyses and a novel measure for functional specificity, we evaluated functional relevance and specificity of topological modules of the human proteome.

CONCLUSIONS

The topological modules ranked using specificity scores show high enrichment with gene sets of known functions. Physical interactions in PPIN contribute to high specificity of the topological modules of the human proteome whereas functional interactions contribute to high homogeneity of the modules. Weighted networks result in more number of topological modules but did not affect their functional propensity. Modules of human proteome are more homogeneous for molecular functions than biological processes.

Concurrence of form and function in developing networks and its role in synaptic pruning

A fundamental question in neuroscience is how structure and function of neural systems are related. We study this interplay by combining a familiar auto-associative neural network with an evolving mechanism for the birth and death of synapses. A feedback loop then arises leading to two qualitatively different types of behaviour. In one, the network structure becomes heterogeneous and dissasortative, and the system displays good memory performance; furthermore, the structure is optimised for the particular memory patterns stored during the process. In the other, the structure remains homogeneous and incapable of pattern retrieval. These findings provide an inspiring picture of brain structure and dynamics that is compatible with experimental results on early brain development, and may help to explain synaptic pruning. Other evolving networks—such as those of protein interactions—might share the basic ingredients for this feedback loop and other questions, and indeed many of their structural features are as predicted by our model.

How structure and function coevolve in developing brains is little understood. Here, the authors study a coupled model of network development and memory, and find that due to the feedback networks with some initial memory capacity evolve into heterogeneous structures with high memory performance.