IsoGloVe: A New Count-based Graph Embedding Method based on Geodesic Distance

Emerging technologies for drug repurposing: Harnessing the potential of text and graph embedding approaches

Advances in proteomic technologies combined with sophisticated computing and modeling methods have generated an unprecedented amount of high-throughput data for system-scale analysis. As a result, the study of protein-protein interaction (PPI) networks has garnered much attention in recent years. One of the most fundamental problems in studying PPI networks is to understand how their architecture originated and evolved to their current state. By investigating how proteins of different ages are connected in the yeast PPI networks, one can deduce their expansion procedure in evolution and how the ancient primitive network expanded and evolved. Studies have shown that proteins are often connected to other proteins of a similar age, suggesting a high degree of age preference between interacting proteins. Though several theories have been proposed to explain this phenomenon, none of them considered protein-clusters as a contributing factor. Here we first investigate the age-dependency of the proteins from the perspective of network motifs. Our analysis confirms that proteins of the same age groups tend to form interacting network motifs; furthermore, those proteins within motifs tend to be within protein complexes and the interactions among them largely contribute to the observed age preference in the yeast PPI networks. In light of these results, we describe a new modeling approach, based on "network motifs", whereby topologically connected protein clusters in the network are treated as single evolutionary units. Instead of modeling single proteins, our approach models the connections and evolutionary relationships of multiple related protein clusters or "network motifs" that are collectively integrated into an existing PPI network. Through simulation studies, we found that the "network motif" modeling approach can capture yeast PPI network properties better than if individual proteins were considered to be the simplest evolutionary units. Our approach provides a fresh perspective on modeling the evolution of yeast PPI networks, specifically that PPI networks may have a much higher age-dependency of interaction density than had been previously envisioned.

We study the question of detecting Conserved Protein Interaction Modules (CPIMs) in protein-protein interaction (PPI) networks. We propose a novel algorithm called GraphHopper that analyzes two PPI networks to find CPIMs. GraphHopper finds CPIMs by “hopping” from one network to another using orthology relationships. By decoupling the degree of evolutionary conservation in a CPIM from the reliability of the PPIs in a CPIM, GraphHopper finds CPIMs with a wide variety of topologies that previous algorithms cannot detect.GraphHopper is competitive with NetworkBlast and Match-and-Split, two state-of-the-art algorithms for computing CPIMs, on the task of recapitulating MIPS processes and complexes. Upon applying GraphHopper to human, fly, and yeast PPI networks, we find a number of CPIMs involved in fundamental processes of the cell that are conserved in all three species. We present the first global map of human-fly CPIMs. This map sheds light on the conservation of protein interaction modules in multi-cellular organisms. CPIMs related to development and the nervous system emerge only in the human-fly comparison. For example, a set of 10 interconnected CPIMs suggest that fly proteins involved in eye development may have human orthologs that have evolved functions related to blood clotting, vascular development, and structural support.KeywordsGene OntologyProtein Interaction NetworkConservation ScoreAlignment GraphIntegrin ComplexThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Proteins are essential macromolecules of life that carry out most cellular processes. Since proteins aggregate to perform function, and since protein-protein interaction (PPI) networks model these aggregations, one would expect to uncover new biology from PPI network topology. Hence, using PPI networks to predict protein function and role of protein pathways in disease has received attention. A debate remains open about whether network properties of “biologically central (BC)” genes (i.e., their protein products), such as those involved in aging, cancer, infectious diseases, or signaling and drug-targeted pathways, exhibit some topological centrality compared to the rest of the proteins in the human PPI network. To help resolve this debate, we design new network-based approaches and apply them to get new insight into biological function and disease. We hypothesize that BC genes have a topologically central (TC) role in the human PPI network. We propose two different concepts of topological centrality. We design a new centrality measure to capture complex wirings of proteins in the network that identifies as TC those proteins that reside in dense extended network neighborhoods. Also, we use the notion of domination and find dominating sets (DSs) in the PPI network, i.e., sets of proteins such that every protein is either in the DS or is a neighbor of the DS. Clearly, a DS has a TC role, as it enables efficient communication between different network parts. We find statistically significant enrichment in BC genes of TC nodes and outperform the existing methods indicating that genes involved in key biological processes occupy topologically complex and dense regions of the network and correspond to its “spine” that connects all other network parts and can thus pass cellular signals efficiently throughout the network. To our knowledge, this is the first study that explores domination in the context of PPI networks.

Intensive research efforts have been undertaken to slow human aging and therefore potentially delay the onset of age-related diseases. These efforts have generated an enormous amount of high-throughput data covering different levels in the physiologic hierarchy, e.g., genetic, epigenetic, transcriptomic, proteomic, and metabolomic, etc. We gathered 15 independent sources of information about genes potentially involved in human longevity and lifespan (N = 5836) and subjected them to various integrated analyses. Many of these genes were initially identified in non-human species, and we investigated their orthologs in three non-human species [i.e., mice (N = 967), fruit fly (N = 449), and worm (N = 411)] for further analysis. We characterized experimentally determined protein-protein interaction networks (PPIN) involving each species’ genes from 9 known protein databases and studied the enriched biological pathways among the individually constructed PPINs. We observed three important signaling pathways: FoxO signaling, mTOR signaling, and autophagy to be common and highly enriched in all four species (p-value ≤ 0.001). Our study implies that the interaction of proteins involved in the mechanistic target of rapamycin (mTOR) signaling pathway is somewhat limited to each species or that a “rewiring” of specific networks has taken place over time. To corroborate our findings, we repeated our analysis in 43 different human tissues. We investigated conserved modules in various tissue-specific PPINs of the longevity-associated genes based upon their protein expression. This analysis also revealed mTOR signaling as shared biological processes across four different human tissue-specific PPINs for liver, heart, skeletal muscle, and adipose tissue. Further, we explored our results’ translational potential by assessing the protein interactions with all the reported drugs and compounds that have been experimentally verified to promote longevity in the three-comparator species. We observed that the target proteins of the FDA-approved drug rapamycin (a known inhibitor of mTOR) were conserved across all four species. Drugs like melatonin and metformin exhibited shared targets with rapamycin in the human PPIN. The detailed information about the curated gene list, cross-species orthologs, PPIN, and pathways was assembled in an interactive data visualization portal using RStudio’s Shiny framework (https://agingnetwork.shinyapps.io/frontiers/).

Protein-protein interaction (PPI) networks provide a well-established framework for elucidating structural properties. In contrast, much greater effort is required to identify useful and/or hidden information using the approach because of the complexity of the networks. Accordingly, an appropriate methodology that allows characterization of a simple, but core, sub-network of a complex network would be a potentially valuable alternative approach. Here, we employed k-core decomposition as a useful tool for obtaining core sub-networks from the complex PPI network in yeast. To apply a k-core value to the complex yeast PPI network, we constructed cytoplasmic and nuclear core sub-networks. We measured the lethality of proteins and the ratio of hub to non-hub core sub-network proteins at each k-core value step. With increasing k-core value, the values of the two parameters gradually increased. Application of 6- and 7-core values to cytoplasmic and nuclear PPI networks yielded modular, structured core sub-networks that were functionally specialized. We also elucidated a false hub: a structural hub exhibiting no functional roles or lethal properties in the network. By employing k-core decomposition to cytoplasmic and nuclear PPI networks in yeast, we obtained useful and/or hidden information during the core sub-network construction process, indicating that k-core decomposition is a useful tool for structural and informational network studies. Notably, we found false hub proteins that might provide insight into protein networks and clues to understanding network evolution.

Perturbation analysis on PCA plus graph embedding methods and PCA plus exponential graph embedding methods

ExEm: Expert embedding using dominating set theory with deep learning approaches

It is well known that protein-protein interaction (PPI) networks are typical evolving complex networks. Identification of important nodes has been an emerging popular topic in complex networks. Many indexes have been proposed to measure the importance of nodes in complex networks, such as degree, closeness, betweenness, k-shell, clustering coefficient, semi-local centrality, eigenvector centrality. Based on multivariate statistical analysis, through integrating the above indexes and further considering the appearances of nodes in network motifs, this paper aims at developing a new measure to characterize the structurally dominant proteins (SDP) in PPI networks. Moreover, we will further investigate the evolution of the defined dominant nodes in temporal evolving real-world and artificial PPI networks. Our results indicate that the constructed artificial networks have some similar statistical properties as those of the real-world evolving networks. In this case, the artificial PPI networks can be used to further investigate the above evolution characteristics of the real-world evolving networks. Simulation results reveal that SDP in the yeast PPI networks are evolutionary conserved, however, the undominant nodes evolve rapidly. Furthermore, PPI networks are very robust against random mutations, while fragile yet with certain robustness to targeted mutations on SDP. Our investigations shed some light on the future applications of the evolving characteristics of bio-molecular networks, such as reengineering of particular networks for technological, synthetic or pharmacological purposes.

Protein–protein interaction (PPI) data derived from biological experiments include many false-positive interactions which are treated equally as other real physical interactions, thereby complicating the detection of real protein complexes from protein–protein interaction (PPI) networks. In this paper, a new weighting method, named as cwMINE (combined weight of module identification in networks), for detecting protein complexes efficiently in protein interaction networks is presented. cwMINE has a good combination between network topology and biological feature, which can solve false positives efficiently of PPI networks and make discovered protein complexes higher quality. In addition, a new expanding rule during the detection process, namely, expanding coefficient, is developed to filter edges with lower weights. The proposed method is compared with several state- of-the-art algorithms in three yeast PPI networks with two benchmark data sets. The experimental results show that the proposed method outperforms the other algorithms in most datasets in terms of the evaluation metrics. We further validate the effectiveness of our method on a human PPI network constructed from the HPRD dataset to identify important disease-related functional modules and provided valuable indications for disease treatment.

In recent years, the analysis of human interaction data has led to the rapid development of graph embedding methods. For link-based classification problems, topological information typically appears in various machine learning tasks in the form of embedded vectors or convolution kernels. This paper introduces a Bayesian graph embedding model for such problems, integrating network reconstruction, link prediction, and behavior prediction into a unified framework. Unlike the existing graph embedding methods, this model does not embed the topology of nodes or links into a low-dimensional space but sorts the probabilities of upcoming links and fuses the information of node topology and data domain via sorting. The new model integrates supervised transaction predictors with unsupervised link prediction models, summarizing local and global topological information. The experimental results on a financial trading dataset and a retweet network dataset demonstrate that the proposed feature fusion model outperforms the tested benchmarked machine learning algorithms in precision, recall, and F1-measure. The proposed learning structure has a fundamental methodological contribution and can be extended and applied to various link-based classification problems in different fields.

The protein-protein interaction (PPI) networks can be regarded as powerful platforms to elucidate the principle and mechanism of cellular organization. Uncovering protein complexes from PPI networks will lead to a better understanding of the science of biological function in cellular systems. In recent decades, numerous computational algorithms have been developed to identify protein complexes. However, the majority of them primarily concern the topological structure of PPI networks and lack of the consideration for the native organized structure among protein complexes. The PPI networks generated by high-throughput technology include a fraction of false protein interactions which make it difficult to identify protein complexes efficiently. To tackle these challenges, we propose a novel semi-supervised protein complex detection model based on non-negative matrix tri-factorization, which not only considers topological structure of a PPI network but also makes full use of available high quality known protein pairs with must-link constraints. We propose non-overlapping (NSSNMTF) and overlapping (OSSNMTF) protein complex detection algorithms to identify the significant protein complexes with clear module structures from PPI networks. In addition, the proposed two protein complex detection algorithms outperform a diverse range of state-of-the-art protein complex identification algorithms on both synthetic networks and human related PPI networks.

BACKGROUNDStudies show that the antifibrotic mechanism of taurine may involve its inhibition of the activation and proliferation of hepatic stellate cells (HSCs). Since the molecular mechanism of taurine-mediated antifibrotic activity has not been fully unveiled and is little studied, it is imperative to use “omics” methods to systematically investigate the molecular mechanism by which taurine inhibits liver fibrosis.AIMTo establish a network including transcriptomic and protein-protein interaction data to elucidate the molecular mechanism of taurine-induced HSC apoptosis.METHODSWe used microarrays, bioinformatics, protein-protein interaction (PPI) network, and sub-modules to investigate taurine-induced changes in gene expression in human HSCs (LX-2). Subsequently, all of the differentially expressed genes (DEGs) were subjected to gene ontology function and Kyoto encyclopedia of genes and genomes pathway enrichment analysis. Furthermore, the interactions of DEGs were explored in a human PPI network, and sub-modules of the DEGs interaction network were analyzed using Cytoscape software.RESULTSA total of 635 DEGs were identified in taurine-treated HSCs when compared with the controls. Of these, 304 genes were statistically significantly up-regulated, and 331 down-regulated. Most of these DEGs were mainly located on the membrane and extracellular region, and are involved in the biological processes of signal transduction, cell proliferation, positive regulation of extracellular regulated protein kinases 1 (ERK1) and ERK2 cascade, extrinsic apoptotic signaling pathway and so on. Fifteen significantly enriched pathways with DEGs were identified, including mitogen-activated protein kinase (MAPK) signaling pathway, peroxisome proliferators-activated receptor signaling pathway, estrogen signaling pathway, Th1 and Th2 cell differentiation, cyclic adenosine monophosphate signaling pathway and so on. By integrating the transcriptomics and human PPI data, nine critical genes, including MMP2, MMP9, MMP21, TIMP3, KLF10, CX3CR1, TGFB1, VEGFB, and EGF, were identified in the PPI network analysis.CONCLUSIONTaurine promotes the apoptosis of HSCs via up-regulating TGFB1 and then activating the p38 MAPK-JNK-Caspase9/8/3 pathway. These findings enhance the understanding of the molecular mechanism of taurine-induced HSC apoptosis and provide references for liver disorder therapy.

Biological functions of a cell are typically carried out through protein complexes. The detection of protein complexes is therefore of great significance for understanding the cellular organizations and protein functions. In the past decades, many computational methods have been proposed to detect protein complexes. However, most of the existing methods just search the local topological information to mine dense subgraphs as protein complexes, ignoring the global topological information. To tackle this issue, we propose the DPCMNE method to detect protein complexes via multi-level network embedding. It can preserve both the local and global topological information of biological networks. First, DPCMNE employs a hierarchical compressing strategy to recursively compress the input protein-protein interaction (PPI) network into multi-level smaller PPI networks. Then, a network embedding method is applied on these smaller PPI networks to learn protein embeddings of different levels of granularity. The embeddings learned from all the compressed PPI networks are concatenated to represent the final protein embeddings of the original input PPI network. Finally, a core-attachment based strategy is adopted to detect protein complexes in the weighted PPI network constructed by the pairwise similarity of protein embeddings. To assess the efficiency of our proposed method, DPCMNE is compared with other eight clustering algorithms on two yeast datasets. The experimental results show that the performance of DPCMNE outperforms those state-of-the-art complex detection methods in terms of F1 and F1+Acc. Furthermore, the results of functional enrichment analysis indicate that protein complexes detected by DPCMNE are more biologically significant in terms of P-score.

Entity Resolution is the process of identifying whether or not various entities from different sources are referring to the same real-world entity. Entity Resolution hasn't been extensively researched in graph databases, whereas it has been for relational databases. This paper focuses on providing comparisons of experiments on various datasets to determine the most appropriate method used in the Entity Resolution process from among literature's similarity algorithms, graph embedding techniques, and graph embedding algorithms combined to link prediction. Moreover, if the embedding algorithm employed has an impact on the given results. The results show that the Entity Resolution process performed better when graph embedding techniques were paired with link prediction, and the chosen graph embedding algorithm also has an impact on the results.