Integrative analysis of human protein, function and disease networks

Predicting protein functions using positive-unlabeled ranking with ontology-based priors

Automated protein function prediction is a crucial and widely studied problem in bioinformatics. Computationally, protein function is a multilabel classification problem where only positive samples are defined and there is a large number of unlabeled annotations. Most existing methods rely on the assumption that the unlabeled set of protein function annotations are negatives, inducing the false negative issue, where potential positive samples are trained as negatives. We introduce a novel approach named PU-GO, wherein we address function prediction as a positive-unlabeled ranking problem. We apply empirical risk minimization, i.e. we minimize the classification risk of a classifier where class priors are obtained from the Gene Ontology hierarchical structure. We show that our approach is more robust than other state-of-the-art methods on similarity-based and time-based benchmark datasets.

AVAILABILITY AND IMPLEMENTATION

Data and code are available at https://github.com/bio-ontology-research-group/PU-GO.

Protein-protein interaction (PPI) network analysis reveals important hub proteins and sub-network modules for root development in rice (Oryza sativa)

BACKGROUND

The root system is vital to plant growth and survival. Therefore, genetic improvement of the root system is beneficial for developing stress-tolerant and improved plant varieties. This requires the identification of proteins that significantly contribute to root development. Analyzing protein-protein interaction (PPI) networks is vastly beneficial in studying developmental phenotypes, such as root development, because a phenotype is an outcome of several interacting proteins. PPI networks can be analyzed to identify modules and get a global understanding of important proteins governing the phenotypes. PPI network analysis for root development in rice has not been performed before and has the potential to yield new findings to improve stress tolerance.

RESULTS

Here, the network module for root development was extracted from the global Oryza sativa PPI network retrieved from the STRING database. Novel protein candidates were predicted, and hub proteins and sub-modules were identified from the extracted module. The validation of the predictions yielded 75 novel candidate proteins, 6 sub-modules, 20 intramodular hubs, and 2 intermodular hubs.

CONCLUSIONS

These results show how the PPI network module is organized for root development and can be used for future wet-lab studies for producing improved rice varieties.

Comparative analysis between high-grade serous ovarian cancer and healthy ovarian tissues using single-cell RNA sequencing

Introduction

High-grade serous ovarian cancer (HGSOC) is the most common histological subtype of ovarian cancer, and is associated with high mortality rates.

Methods

In this study, we analyzed specific cell subpopulations and compared different gene functions between healthy ovarian and ovarian cancer cells using single-cell RNA sequencing (ScRNA-seq). We delved deeper into the differences between healthy ovarian and ovarian cancer cells at different levels, and performed specific analysis on endothelial cells.

Results

We obtained scRNA-seq data of 6867 and 17056 cells from healthy ovarian samples and ovarian cancer samples, respectively. The transcriptional profiles of the groups differed at various stages of ovarian cell development. A detailed comparison of the cell cycle, and cell communication of different groups, revealed significant differences between healthy ovarian and ovarian cancer cells. We also found that apoptosis-related genes, URI1, PAK2, PARP1, CLU and TIMP3, were highly expressed, while immune-related genes, UBB, RPL11, CAV1, NUPR1 and Hsp90ab1, were lowly expressed in ovarian cancer cells. The results of the ScRNA-seq were verified using qPCR.

Discussion

Our findings revealed differences in function, gene expression and cell interaction patterns between ovarian cancer and healthy ovarian cell populations. These findings provide key insights on further research into the treatment of ovarian cancer.

A fingerprint of plasma proteome alteration after local tissue damage induced by Bothrops leucurus snake venom in mice

Bothrops spp. is responsible for about 70% of snakebites in Brazil, causing a diverse and complex pathophysiological condition. Bothrops leucurus is the main species of medical relevance found in the Atlantic coast in the Brazilian Northeast region. The pathophysiological effects involved B. leucurus snakebite as well as the organism's reaction in response to this envenoming, it has not been explored yet. Thus, edema was induced in mice paw using 1.2, 2.5, and 5.0 μg of B. leucurus venom, the percentage of edema was measured 30 min after injection and the blood plasma was collected and analyzed by shotgun proteomic strategy. We identified 80 common plasma proteins with differential abundance among the experimental groups and we can understand the early aspects of this snake envenomation, regardless of the suggestive severity of an ophidian accident. The results showed B. leucurus venom triggers a thromboinflammation scenario where family's proteins of the Serpins, Apolipoproteins, Complement factors and Component subunits, Cathepsins, Kinases, Oxidoreductases, Proteases inhibitors, Proteases, Collagens, Growth factors are related to inflammation, complement and coagulation systems, modulators platelets and neutrophils, lipid and retinoid metabolism, oxidative stress and tissue repair. Our findings set precedents for future studies in the area of early diagnosis and/or treatment of snakebites. SIGNIFICANCE: The physiopathological effects that the snake venoms can cause have been investigated through classical and reductionist tools, which allowed, so far, the identification of action mechanisms of individual components associated with specific tissue damage. The currently incomplete limitations of this knowledge must be expanded through new approaches, such as proteomics, which may represent a big leap in understanding the venom-modulated pathological process. The exploration of the complete protein set that suffer modifications by the simultaneous action of multiple toxins, provides a map of the establishment of physiopathological phenotypes, which favors the identification of multiple toxin targets, that may or may not act in synergy, as well as favoring the discovery of biomarkers and therapeutic targets for manifestations that are not neutralized by the antivenom.

Whole-genome sequencing reveals novel ethnicity-specific rare variants associated with Alzheimer's disease

Alzheimer's disease (AD) is the most common multifactorial neurodegenerative disease among elderly people. Genome-wide association studies (GWAS) have been highly successful in identifying genetic risk factors. However, GWAS investigate common variants, which tend to have small effect sizes, and rare variants with potentially larger phenotypic effects have not been sufficiently investigated. Whole-genome sequencing (WGS) enables us to detect those rare variants. Here, we performed rare-variant association studies by using WGS data from 140 individuals with probable AD and 798 cognitively normal elder controls (CN), as well as single-nucleotide polymorphism genotyping data from an independent large Japanese AD cohort of 1604 AD and 1235 CN subjects. We identified two rare variants as candidates for AD association: a missense variant in OR51G1 (rs146006146, c.815 G > A, p.R272H) and a stop-gain variant in MLKL (rs763812068, c.142 C > T, p.Q48X). Subsequent in vitro functional analysis revealed that the MLKL stop-gain variant can contribute to increases not only in abnormal cells that should die by programmed cell death but do not, but also in the ratio of Aβ42 to Aβ40. We further detected AD candidate genes through gene-based association tests of rare variants; a network-based meta-analysis using these candidates identified four functionally important hub genes (NCOR2, PLEC, DMD, and NEDD4). Our findings will contribute to the understanding of AD and provide novel insights into its pathogenic mechanisms that can be used in future studies.

Exploring the conservation of Alzheimer-related pathways between H. sapiens and C. elegans: a network alignment approach

Alzheimer disease (AD) is a neurodegenerative disorder with an –as of yet– unclear etiology and pathogenesis. Research to unveil disease processes underlying AD often relies on the use of neurodegenerative disease model organisms, such as Caenorhabditis elegans. This study sought to identify biological pathways implicated in AD that are conserved in Homo sapiens and C. elegans. Protein–protein interaction networks were assembled for amyloid precursor protein (APP) and Tau in H. sapiens—two proteins whose aggregation is a hallmark in AD—and their orthologs APL-1 and PTL-1 for C. elegans. Global network alignment was used to compare these networks and determine similar, likely conserved, network regions. This comparison revealed that two prominent pathways, the APP-processing and the Tau-phosphorylation pathways, are highly conserved in both organisms. While the majority of interactions between proteins in those pathways are known to be associated with AD in human, they remain unexamined in C. elegans, signifying the need for their further investigation. In this work, we have highlighted conserved interactions related to AD in humans and have identified specific proteins that can act as targets for experimental studies in C. elegans, aiming to uncover the underlying mechanisms of AD.

Statistical Physics for Medical Diagnostics: Learning, Inference, and Optimization Algorithms

It is widely believed that cooperation between clinicians and machines may address many of the decisional fragilities intrinsic to current medical practice. However, the realization of this potential will require more precise definitions of disease states as well as their dynamics and interactions. A careful probabilistic examination of symptoms and signs, including the molecular profiles of the relevant biochemical networks, will often be required for building an unbiased and efficient diagnostic approach. Analogous problems have been studied for years by physicists extracting macroscopic states of various physical systems by examining microscopic elements and their interactions. These valuable experiences are now being extended to the medical field. From this perspective, we discuss how recent developments in statistical physics, machine learning and inference algorithms are coming together to improve current medical diagnostic approaches.

Disease evolution in reaction networks: Implications for a diagnostic problem

We study the time evolution of symptoms (signs) with some defects in the dynamics of a reaction network as a (microscopic) model for the progress of disease phenotypes. To this end, we take a large population of reaction networks and follow the stochastic dynamics of the system to see how the development of defects affects the macroscopic states of the signs probability distribution. We start from some plausible definitions for the healthy and disease states along with a dynamical model for the emergence of diseases by a reverse simulated annealing algorithm. The healthy state is defined as a state of maximum objective function, which here is the sum of mutual information between a subset of signal variables and the subset of assigned response variables. A disease phenotype is defined with two parameters controlling the rate of mutations in reactions and the rate of accepting mutations that reduce the objective function. The model can provide the time dependence of the sign probabilities given a disease phenotype. This allows us to obtain the accuracy of diagnosis as a function of time by using a probabilistic model of signs and diseases. The trade-off between the diagnosis accuracy (increasing in time) and the objective function (decreasing in time) can be used to suggest an optimal time for medical intervention. Our model would be useful in particular for a dynamical (history-based) diagnostic problem, to estimate the likelihood of a disease hypothesis given the temporal evolution of the signs.

Here, we use concepts from statistical physics and reaction network dynamics to introduce a measure to quantify the tradeoff between the accuracy of diagnosis and an early diagnosis. This measure is used to suggest an optimal time for medical intervention depending on the number of observed signs (medical tests). We present a stochastic model using a reverse simulated annealing algorithm for numerical simulation of disease evolution. The model can provide the time dependence of the sign probabilities given a disease phenotype. This in turn allows us to anticipate the accuracy of diagnosis as a function of time by using a probabilistic model of signs and diseases.

Graph cluster approach in identifying novel proteins and significant pathways involved in polycystic ovary syndrome

RESEARCH QUESTION

Polycystic ovary syndrome (PCOS) is a complex endocrine disorder with diverse clinical implications, such as infertility, metabolic disorders, cardiovascular diseases and psychological problems among others. The heterogeneity of conditions found in PCOS contribute to its various phenotypes, leading to difficulties in identifying proteins involved in this abnormality. Several studies, however, have shown the feasibility in identifying molecular evidence underlying other diseases using graph cluster analysis. Therefore, is it possible to identify proteins and pathways related to PCOS using the same approach?

METHODS

Known PCOS-related proteins (PCOSrp) from PCOSBase and DisGeNET were integrated with protein-protein interactions (PPI) information from Human Integrated Protein-Protein Interaction reference to construct a PCOS PPI network. The network was clustered with DPClusO algorithm to generate clusters, which were evaluated using Fisher's exact test. Pathway enrichment analysis using gProfileR was conducted to identify significant pathways.

RESULTS

The statistical significance of the identified clusters has successfully predicted 138 novel PCOSrp with 61.5% reliability and, based on Cronbach's alpha, this prediction is acceptable. Androgen signalling pathway and leptin signalling pathway were among the significant PCOS-related pathways corroborating the information obtained from the clinical observation, where androgen signalling pathway is responsible in producing male hormones in women with PCOS, whereas leptin signalling pathway is involved in insulin sensitivity.

CONCLUSIONS

These results show that graph cluster analysis can provide additional insight into the pathobiology of PCOS, as the pathways identified as statistically significant correspond to earlier biological studies. Therefore, integrative analysis can reveal unknown mechanisms, which may enable the development of accurate diagnosis and effective treatment in PCOS.

Genetic and Expression Analysis of COPI Genes and Alzheimer's Disease Susceptibility

Alzheimer's disease (AD) is the most common neurodegenerative disease in the elderly and the leading cause of dementia in humans. Evidence shows that cellular trafficking and recycling machineries are associated with AD risk. A recent study found that the coat protein complex I (COPI)-dependent trafficking in vivo could significantly reduce amyloid plaques in the cortex and hippocampus of neurological in the AD mouse models and identified 12 single-nucleotide polymorphisms in COPI genes to be significantly associated with increased AD risk using 6,795 samples. Here, we used a large-scale GWAS dataset to investigate the potential association between the COPI genes and AD susceptibility by both SNP and gene-based tests. The results showed that only rs9898218 was associated with AD risk with P = 0.017. We further conducted an expression quantitative trait loci (eQTLs) analysis and found that rs9898218 G allele was associated with increased COPZ2 expression in cerebellar cortex with P = 0.0184. Importantly, the eQTLs analysis in whole blood further indicated that 11 of these 12 genetic variants could significantly regulate the expression of COPI genes. Hence, these findings may contribute to understand the association between COPI genes and AD susceptibility.

Analysis of protein targets in pathogen-host interaction in infectious diseases: a case study on Plasmodium falciparum and Homo sapiens interaction network

Infection and disease progression is the outcome of protein interactions between pathogen and host. Pathogen, the role player of Infection, is becoming a severe threat to life as because of its adaptability toward drugs and evolutionary dynamism in nature. Identifying protein targets by analyzing protein interactions between host and pathogen is the key point. Proteins with higher degree and possessing some topologically significant graph theoretical measures are found to be drug targets. On the other hand, exceptional nodes may be involved in infection mechanism because of some pathway process and biologically unknown factors. In this article, we attempt to investigate characteristics of host-pathogen protein interactions by presenting a comprehensive review of computational approaches applied on different infectious diseases. As an illustration, we have analyzed a case study on infectious disease malaria, with its causative agent Plasmodium falciparum acting as 'Bait' and host, Homo sapiens/human acting as 'Prey'. In this pathogen-host interaction network based on some interconnectivity and centrality properties, proteins are viewed as central, peripheral, hub and non-hub nodes and their significance on infection process. Besides, it is observed that because of sparseness of the pathogen and host interaction network, there may be some topologically unimportant but biologically significant proteins, which can also act as Bait/Prey. So, functional similarity or gene ontology mapping can help us in this case to identify these proteins.

Disease association of human tumor suppressor genes

The multifactorial disease, cancer, frequently emerges due to perturbations in tumor suppressor genes (TSGs). However, a growing number of noncanonical target genes of TSGs and the highly interconnected nature of the human interactome reveal that the functions of TSGs are not limited to cancer-specific events. The various functions of TSGs lead to the assumption that cancer is linked with other human disorders. Therefore, a disease-gene association network of TSGs (TSDN) was constructed by integrating protein-protein interaction networks of TSGs (TSN) with Morbid Map in Online Mendelian Inheritance in Man. The TSDN revealed links between TSGs and 22 different human disorders including cancer and indicated disease-disease associations. In addition, high-density functional protein clusters in the TSN showed cohesive and overlapping disease-TSG associations, which proved the prevalent role of TSGs in various human diseases beyond cancer. The presence of overlapping disease-gene modules and disease-disease associations via the TSN demonstrated that other diseases can serve as possible roots of the life-threatening disease cancer. Therefore, a disease association map of TSGs could be a promising tool for exploring intricate relationships between cancer and other diseases for the early prediction of cancer and the understanding of disease etiology.

Mapping biological process relationships and disease perturbations within a pathway network

Molecular interaction networks are routinely used to map the organization of cellular function. Edges represent interactions between genes, proteins, or metabolites. However, in living cells, molecular interactions are dynamic, necessitating context-dependent models. Contextual information can be integrated into molecular interaction networks through the inclusion of additional molecular data, but there are concerns about completeness and relevance of this data. We developed an approach for representing the organization of human cellular processes using pathways as the nodes in a network. Pathways represent spatial and temporal sets of context-dependent interactions, generating a high-level network when linked together, which incorporates contextual information without the need for molecular interaction data. Analysis of the pathway network revealed linked communities representing functional relationships, comparable to those found in molecular networks, including metabolism, signaling, immunity, and the cell cycle. We mapped a range of diseases onto this network and find that pathways associated with diseases tend to be functionally connected, highlighting the perturbed functions that result in disease phenotypes. We demonstrated that disease pathways cluster within the network. We then examined the distribution of cancer pathways and showed that cancer pathways tend to localize within the signaling, DNA processes and immune modules, although some cancer-associated nodes are found in other network regions. Altogether, we generated a high-confidence functional network, which avoids some of the shortcomings faced by conventional molecular models. Our representation provides an intuitive functional interpretation of cellular organization, which relies only on high-quality pathway and Gene Ontology data. The network is available at https://data.mendeley.com/datasets/3pbwkxjxg9/1.

Statistical physics of medical diagnostics: Study of a probabilistic model

We study a diagnostic strategy which is based on the anticipation of the diagnostic process by simulation of the dynamical process starting from the initial findings. We show that such a strategy could result in more accurate diagnoses compared to a strategy that is solely based on the direct implications of the initial observations. We demonstrate this by employing the mean-field approximation of statistical physics to compute the posterior disease probabilities for a given subset of observed signs (symptoms) in a probabilistic model of signs and diseases. A Monte Carlo optimization algorithm is then used to maximize an objective function of the sequence of observations, which favors the more decisive observations resulting in more polarized disease probabilities. We see how the observed signs change the nature of the macroscopic (Gibbs) states of the sign and disease probability distributions. The structure of these macroscopic states in the configuration space of the variables affects the quality of any approximate inference algorithm (so the diagnostic performance) which tries to estimate the sign-disease marginal probabilities. In particular, we find that the simulation (or extrapolation) of the diagnostic process is helpful when the disease landscape is not trivial and the system undergoes a phase transition to an ordered phase.

Disease Risk Assessment Using a Voronoi-Based Network Analysis of Genes and Variants Scores

Much effort has been devoted to assess disease risk based on large-scale protein-protein network and genotype-phenotype associations. However, the challenge of risk prediction for complex diseases remains unaddressed. Here, we propose a framework to quantify the risk based on a Voronoi tessellation network analysis, taking into account the disease association scores of both genes and variants. By integrating ClinVar, SNPnexus, and DISEASES databases, we introduce a gene-variant map that is based on the pairwise disease-associated gene-variant scores. This map is clustered using Voronoi tessellation and network analysis with a threshold obtained from fitting the background Voronoi cell density distribution. We define the relative risk of disease that is inferred from the scores of the data points within the related clusters on the gene-variant map. We identify autoimmune-associated clusters that may interact at the system-level. The proposed framework can be used to determine the clusters that are specific to a subtype or contribute to multiple subtypes of complex diseases.