Predicting gene disease associations with knowledge graph embeddings for diseases with curtailed information.

Knowledge graph embeddings (KGE) are a powerful technique used in the biomedical domain to represent biological knowledge in a low dimensional space. However, a deep understanding of these methods is still missing, and, in particular, regarding their applications to prioritize genes associated with complex diseases with reduced genetic information. In this contribution, we built a knowledge graph (KG) by integrating heterogeneous biomedical data and generated KGE by implementing state-of-the-art methods, and two novel algorithms: Dlemb and BioKG2vec. Extensive testing of the embeddings with unsupervised clustering and supervised methods showed that KGE can be successfully implemented to predict genes associated with diseases and that our novel approaches outperform most existing algorithms in both scenarios. Our findings underscore the significance of data quality, preprocessing, and integration in achieving accurate predictions. Additionally, we applied KGE to predict genes linked to Intervertebral Disc Degeneration (IDD) and illustrated that functions pertinent to the disease are enriched within the prioritized gene set.

A dynamic model of the intestinal epithelium integrates multiple sources of preclinical data and enables clinical translation of drug-induced toxicity.

We have built a quantitative systems toxicology modeling framework focused on the early prediction of oncotherapeutic-induced clinical intestinal adverse effects. The model describes stem and progenitor cell dynamics in the small intestinal epithelium and integrates heterogeneous epithelial-related processes, such as transcriptional profiles, citrulline kinetics, and probability of diarrhea. We fitted a mouse-specific version of the model to quantify doxorubicin and 5-fluorouracil (5-FU)-induced toxicity, which included pharmacokinetics and 5-FU metabolism and assumed that both drugs led to cell cycle arrest and apoptosis in stem cells and proliferative progenitors. The model successfully recapitulated observations in mice regarding dose-dependent disruption of proliferation which could lead to villus shortening, decrease of circulating citrulline, increased diarrhea risk, and transcriptional induction of the p53 pathway. Using a human-specific epithelial model, we translated the cytotoxic activity of doxorubicin and 5-FU quantified in mice into human intestinal injury and predicted with accuracy clinical diarrhea incidence. However, for gefitinib, a specific-molecularly targeted therapy, the mice failed to reproduce epithelial toxicity at exposures much higher than those associated with clinical diarrhea. This indicates that, regardless of the translational modeling approach, preclinical experimental settings have to be suitable to quantify drug-induced clinical toxicity with precision at the structural scale of the model. Our work demonstrates the usefulness of translational models at early stages of the drug development pipeline to predict clinical toxicity and highlights the importance of understanding cross-settings differences in toxicity when building these approaches.

Assessing network-based methods in the context of system toxicology.

Introduction: Network-based methods are promising approaches in systems toxicology because they can be used to predict the effects of drugs and chemicals on health, to elucidate the mode of action of compounds, and to identify biomarkers of toxicity. Over the years, the network biology community has developed a wide range of methods, and users are faced with the task of choosing the most appropriate method for their own application. Furthermore, the advantages and limitations of each method are difficult to determine without a proper standard and comparative evaluation of their performance. This study aims to evaluate different network-based methods that can be used to gain biological insight into the mechanisms of drug toxicity, using valproic acid (VPA)-induced liver steatosis as a benchmark. Methods: We provide a comprehensive analysis of the results produced by each method and highlight the fact that the experimental design (how the method is applied) is relevant in addition to the method specifications. We also contribute with a systematic methodology to analyse the results of the methods individually and in a comparative manner. Results: Our results show that the evaluated tools differ in their performance against the benchmark and in their ability to provide novel insights into the mechanism of adverse effects of the drug. We also suggest that aggregation of the results provided by different methods provides a more confident set of candidate genes and processes to further the knowledge of the drug's mechanism of action. Discussion: By providing a detailed and systematic analysis of the results of different network-based tools, we aim to assist users in making informed decisions about the most appropriate method for systems toxicology applications.

Visualization of automatically combined disease maps and pathway diagrams for rare diseases.

Introduction: Investigation of molecular mechanisms of human disorders, especially rare diseases, require exploration of various knowledge repositories for building precise hypotheses and complex data interpretation. Recently, increasingly more resources offer diagrammatic representation of such mechanisms, including disease-dedicated schematics in pathway databases and disease maps. However, collection of knowledge across them is challenging, especially for research projects with limited manpower. Methods: In this article we present an automated workflow for construction of maps of molecular mechanisms for rare diseases. The workflow requires a standardized definition of a disease using Orphanet or HPO identifiers to collect relevant genes and variants, and to assemble a functional, visual repository of related mechanisms, including data overlays. The diagrams composing the final map are unified to a common systems biology format from CellDesigner SBML, GPML and SBML+layout+render. The constructed resource contains disease-relevant genes and variants as data overlays for immediate visual exploration, including embedded genetic variant browser and protein structure viewer. Results: We demonstrate the functionality of our workflow on two examples of rare diseases: Kawasaki disease and retinitis pigmentosa. Two maps are constructed based on their corresponding identifiers. Moreover, for the retinitis pigmentosa use-case, we include a list of differentially expressed genes to demonstrate how to tailor the workflow using omics datasets. Discussion: In summary, our work allows for an ad-hoc construction of molecular diagrams combined from different sources, preserving their layout and graphical style, but integrating them into a single resource. This allows to reduce time consuming tasks of prototyping of a molecular disease map, enabling visual exploration, hypothesis building, data visualization and further refinement. The code of the workflow is open and accessible at https://gitlab.lcsb.uni.lu/minerva/automap/.

Genomic and proteomic biomarker landscape in clinical trials.

The use of molecular biomarkers to support disease diagnosis, monitor its progression, and guide drug treatment has gained traction in the last decades. While only a dozen biomarkers have been approved for their exploitation in the clinic by the FDA, many more are evaluated in the context of translational research and clinical trials. Furthermore, the information on which biomarkers are measured, for which purpose, and in relation to which conditions are not readily accessible: biomarkers used in clinical studies available through resources such as ClinicalTrials.gov are described as free text, posing significant challenges in finding, analyzing, and processing them by both humans and machines. We present a text mining strategy to identify proteomic and genomic biomarkers used in clinical trials and classify them according to the methodologies by which they are measured. We find more than 3000 biomarkers used in the context of 2600 diseases. By analyzing this dataset, we uncover patterns of use of biomarkers across therapeutic areas over time, including the biomarker type and their specificity. These data are made available at the Clinical Biomarker App at https://www.disgenet.org/biomarkers/, a new portal that enables the exploration of biomarkers extracted from the clinical studies available at ClinicalTrials.gov and enriched with information from the scientific literature. The App features several metrics that assess the specificity of the biomarkers, facilitating their selection and prioritization. Overall, the Clinical Biomarker App is a valuable and timely resource about clinical biomarkers, to accelerate biomarker discovery, development, and application.

Drug-target identification in COVID-19 disease mechanisms using computational systems biology approaches.

Introduction: The COVID-19 Disease Map project is a large-scale community effort uniting 277 scientists from 130 Institutions around the globe. We use high-quality, mechanistic content describing SARS-CoV-2-host interactions and develop interoperable bioinformatic pipelines for novel target identification and drug repurposing. Methods: Extensive community work allowed an impressive step forward in building interfaces between Systems Biology tools and platforms. Our framework can link biomolecules from omics data analysis and computational modelling to dysregulated pathways in a cell-, tissue- or patient-specific manner. Drug repurposing using text mining and AI-assisted analysis identified potential drugs, chemicals and microRNAs that could target the identified key factors. Results: Results revealed drugs already tested for anti-COVID-19 efficacy, providing a mechanistic context for their mode of action, and drugs already in clinical trials for treating other diseases, never tested against COVID-19. Discussion: The key advance is that the proposed framework is versatile and expandable, offering a significant upgrade in the arsenal for virus-host interactions and other complex pathologies.

Benchmarking post-GWAS analysis tools in major depression: Challenges and implications.

Our knowledge of complex disorders has increased in the last years thanks to the identification of genetic variants (GVs) significantly associated with disease phenotypes by genome-wide association studies (GWAS). However, we do not understand yet how these GVs functionally impact disease pathogenesis or their underlying biological mechanisms. Among the multiple post-GWAS methods available, fine-mapping and colocalization approaches are commonly used to identify causal GVs, meaning those with a biological effect on the trait, and their functional effects. Despite the variety of post-GWAS tools available, there is no guideline for method eligibility or validity, even though these methods work under different assumptions when accounting for linkage disequilibrium and integrating molecular annotation data. Moreover, there is no benchmarking of the available tools. In this context, we have applied two different fine-mapping and colocalization methods to the same GWAS on major depression (MD) and expression quantitative trait loci (eQTL) datasets. Our goal is to perform a systematic comparison of the results obtained by the different tools. To that end, we have evaluated their results at different levels: fine-mapped and colocalizing GVs, their target genes and tissue specificity according to gene expression information, as well as the biological processes in which they are involved. Our findings highlight the importance of fine-mapping as a key step for subsequent analysis. Notably, the colocalizing variants, altered genes and targeted tissues differed between methods, even regarding their biological implications. This contribution illustrates an important issue in post-GWAS analysis with relevant consequences on the use of GWAS results for elucidation of disease pathobiology, drug target prioritization and biomarker discovery.

Functional Genomics Analysis to Disentangle the Role of Genetic Variants in Major Depression.

Understanding the molecular basis of major depression is critical for identifying new potential biomarkers and drug targets to alleviate its burden on society. Leveraging available GWAS data and functional genomic tools to assess regulatory variation could help explain the role of major depression-associated genetic variants in disease pathogenesis. We have conducted a fine-mapping analysis of genetic variants associated with major depression and applied a pipeline focused on gene expression regulation by using two complementary approaches: cis-eQTL colocalization analysis and alteration of transcription factor binding sites. The fine-mapping process uncovered putative causally associated variants whose proximal genes were linked with major depression pathophysiology. Four colocalizing genetic variants altered the expression of five genes, highlighting the role of SLC12A5 in neuronal chlorine homeostasis and MYRF in nervous system myelination and oligodendrocyte differentiation. The transcription factor binding analysis revealed the potential role of rs62259947 in modulating P4HTM expression by altering the YY1 binding site, altogether regulating hypoxia response. Overall, our pipeline could prioritize putative causal genetic variants in major depression. More importantly, it can be applied when only index genetic variants are available. Finally, the presented approach enabled the proposal of mechanistic hypotheses of these genetic variants and their role in disease pathogenesis.

SARS-CoV-2 sculpts the immune system to induce sustained virus-specific naïve-like and memory B-cell responses.

OBJECTIVES: SARS-CoV-2 infection induces virus-reactive memory B cells expressing unmutated antibodies, which hints at their emergence from naïve B cells. Yet, the dynamics of virus-specific naïve B cells and their impact on immunity and immunopathology remain unclear. METHODS: We longitudinally profiled SARS-CoV-2-specific B-cell responses in 25 moderate-to-severe COVID-19 patients by high-dimensional flow cytometry and isotyping and subtyping ELISA. We also explored the relationship of B-cell responses to SARS-CoV-2 with the activation of effector and regulatory cells from the innate or adaptive immune system. RESULTS: We found a virus-specific antibody response with a broad spectrum of classes and subclasses during acute infection, which evolved into an IgG1-dominated response during convalescence. Acute infection was associated with increased mature B-cell progenitors in the circulation and the unexpected expansion of virus-targeting naïve-like B cells. The latter further augmented during convalescence together with virus-specific memory B cells. In addition to a transitory increase in tissue-homing CXCR3+ plasmablasts and extrafollicular memory B cells, most COVID-19 patients showed persistent activation of CD4+ and CD8+ T cells along with transient or long-lasting changes of key innate immune cells. Remarkably, virus-specific antibodies and the frequency of naïve B cells were among the major variables defining distinct immune signatures associated with disease severity and inflammation. CONCLUSION: Aside from providing new insights into the complexity of the immune response to SARS-CoV-2, our findings indicate that the de novo recruitment of mature B-cell precursors into the periphery may be central to the induction of antiviral immunity.

The DisGeNET cytoscape app: Exploring and visualizing disease genomics data.

Thanks to the unbiased exploration of genomic variants at large scale, hundreds of thousands of disease-associated loci have been uncovered. In parallel, network-based approaches have proven to be essential to understand the molecular mechanisms underlying human diseases. The use of these approaches has been boosted by the abundance of information about disease associated genes and variants, high quality human interactomics data, and the emergence of new types of omics data. The DisGeNET Cytoscape App combines the capabilities of Cytoscape with those of DisGeNET, a knowledge platform based on a comprehensive catalogue of disease-associated genes and variants. The DisGeNET Cytoscape App contains functions to query, analyze, and visualize different network representations of the gene-disease and variant-disease associations available in DisGeNET. It supports a wide variety of applications through its query and filter functionalities, including the annotation of foreign networks generated by other apps or uploaded by the user. The new release of the DisGeNET Cytoscape App has been designed to support Cytoscape 3.x and incorporates novel distinctive features such as visualization and analysis of variant-disease networks, disease enrichment analysis for genes and variants, and analytic support through Cytoscape Automation. Moreover, the DisGeNET Cytoscape App features an API to access its core functionalities via the REST protocol fostering the development of reproducible and scalable analysis workflows based on DisGeNET data.

The eTRANSAFE Project on Translational Safety Assessment through Integrative Knowledge Management: Achievements and Perspectives.

eTRANSAFE is a research project funded within the Innovative Medicines Initiative (IMI), which aims at developing integrated databases and computational tools (the eTRANSAFE ToxHub) that support the translational safety assessment of new drugs by using legacy data provided by the pharmaceutical companies that participate in the project. The project objectives include the development of databases containing preclinical and clinical data, computational systems for translational analysis including tools for data query, analysis and visualization, as well as computational models to explain and predict drug safety events.

Comorbidity between Alzheimer's disease and major depression: a behavioural and transcriptomic characterization study in mice.

BACKGROUND: Major depression (MD) is the most prevalent psychiatric disease in the population and is considered a prodromal stage of the Alzheimer's disease (AD). Despite both diseases having a robust genetic component, the common transcriptomic signature remains unknown. METHODS: We investigated the cognitive and emotional behavioural responses in 3- and 6-month-old APP/PSEN1-Tg mice, before ß-amyloid plaques were detected. We studied the genetic and pathway deregulation in the prefrontal cortex, striatum, hippocampus and amygdala of mice at both ages, using transcriptomic and functional data analysis. RESULTS: We found that depressive-like and anxiety-like behaviours, as well as memory impairments, are already present at 3-month-old APP/PSEN1-Tg mutant mice together with the deregulation of several genes, such as Ciart, Grin3b, Nr1d1 and Mc4r, and other genes including components of the circadian rhythms, electron transport chain and neurotransmission in all brain areas. Extending these results to human data performing GSEA analysis using DisGeNET database, it provides translational support for common deregulated gene sets related to MD and AD. CONCLUSIONS: The present study sheds light on the shared genetic bases between MD and AD, based on a comprehensive characterization from the behavioural to transcriptomic level. These findings suggest that late MD could be an early manifestation of AD.

An ensemble learning approach for modeling the systems biology of drug-induced injury.

BACKGROUND: Drug-induced liver injury (DILI) is an adverse reaction caused by the intake of drugs of common use that produces liver damage. The impact of DILI is estimated to affect around 20 in 100,000 inhabitants worldwide each year. Despite being one of the main causes of liver failure, the pathophysiology and mechanisms of DILI are poorly understood. In the present study, we developed an ensemble learning approach based on different features (CMap gene expression, chemical structures, drug targets) to predict drugs that might cause DILI and gain a better understanding of the mechanisms linked to the adverse reaction. RESULTS: We searched for gene signatures in CMap gene expression data by using two approaches: phenotype-gene associations data from DisGeNET, and a non-parametric test comparing gene expression of DILI-Concern and No-DILI-Concern drugs (as per DILIrank definitions). The average accuracy of the classifiers in both approaches was 69%. We used chemical structures as features, obtaining an accuracy of 65%. The combination of both types of features produced an accuracy around 63%, but improved the independent hold-out test up to 67%. The use of drug-target associations as feature obtained the best accuracy (70%) in the independent hold-out test. CONCLUSIONS: When using CMap gene expression data, searching for a specific gene signature among the landmark genes improves the quality of the classifiers, but it is still limited by the intrinsic noise of the dataset. When using chemical structures as a feature, the structural diversity of the known DILI-causing drugs hampers the prediction, which is a similar problem as for the use of gene expression information. The combination of both features did not improve the quality of the classifiers but increased the robustness as shown on independent hold-out tests. The use of drug-target associations as feature improved the prediction, specially the specificity, and the results were comparable to previous research studies.

The ELIXIR Human Copy Number Variations Community: building bioinformatics infrastructure for research.

Copy number variations (CNVs) are major causative contributors both in the genesis of genetic diseases and human neoplasias. While "High-Throughput" sequencing technologies are increasingly becoming the primary choice for genomic screening analysis, their ability to efficiently detect CNVs is still heterogeneous and remains to be developed. The aim of this white paper is to provide a guiding framework for the future contributions of ELIXIR's recently established human CNV Community, with implications beyond human disease diagnostics and population genomics. This white paper is the direct result of a strategy meeting that took place in September 2018 in Hinxton (UK) and involved representatives of 11 ELIXIR Nodes. The meeting led to the definition of priority objectives and tasks, to address a wide range of CNV-related challenges ranging from detection and interpretation to sharing and training. Here, we provide suggestions on how to align these tasks within the ELIXIR Platforms strategy, and on how to frame the activities of this new ELIXIR Community in the international context.

The DisGeNET knowledge platform for disease genomics: 2019 update.

One of the most pressing challenges in genomic medicine is to understand the role played by genetic variation in health and disease. Thanks to the exploration of genomic variants at large scale, hundreds of thousands of disease-associated loci have been uncovered. However, the identification of variants of clinical relevance is a significant challenge that requires comprehensive interrogation of previous knowledge and linkage to new experimental results. To assist in this complex task, we created DisGeNET (http://www.disgenet.org/), a knowledge management platform integrating and standardizing data about disease associated genes and variants from multiple sources, including the scientific literature. DisGeNET covers the full spectrum of human diseases as well as normal and abnormal traits. The current release covers more than 24 000 diseases and traits, 17 000 genes and 117 000 genomic variants. The latest developments of DisGeNET include new sources of data, novel data attributes and prioritization metrics, a redesigned web interface and recently launched APIs. Thanks to the data standardization, the combination of expert curated information with data automatically mined from the scientific literature, and a suite of tools for accessing its publicly available data, DisGeNET is an interoperable resource supporting a variety of applications in genomic medicine and drug R&D.

ResMarkerDB: a database of biomarkers of response to antibody therapy in breast and colorectal cancer.

The clinical efficacy of therapeutic monoclonal antibodies for breast and colorectal cancer has greatly contributed to the improvement of patients' outcomes by individualizing their treatments according to their genomic background. However, primary or acquired resistance to treatment reduces its efficacy. In this context, the identification of biomarkers predictive of drug response would support research and development of new alternative treatments. Biomarkers play a major role in the genomic revolution, supporting disease diagnosis and treatment decision-making. Currently, several molecular biomarkers of treatment response for breast and colorectal cancer have been described. However, information on these biomarkers is scattered across several resources, and needs to be identified, collected and properly integrated to be fully exploited to inform monitoring of drug response in patients. Therefore, there is a need of resources that offer biomarker data in a harmonized manner to the user to support the identification of actionable biomarkers of response to treatment in cancer. ResMarkerDB was developed as a comprehensive resource of biomarkers of drug response in colorectal and breast cancer. It integrates data of biomarkers of drug response from existing repositories, and new data extracted and curated from the literature (referred as ResCur). ResMarkerDB currently features 266 biomarkers of diverse nature. Twenty-five percent of these biomarkers are exclusive of ResMarkerDB. Furthermore, ResMarkerDB is one of the few resources offering non-coding DNA data in response to drug treatment. The database contains more than 500 biomarker-drug-tumour associations, covering more than 100 genes. ResMarkerDB provides a web interface to facilitate the exploration of the current knowledge of biomarkers of response in breast and colorectal cancer. It aims to enhance translational research efforts in identifying actionable biomarkers of drug response in cancer.

GUILDify v2.0: A Tool to Identify Molecular Networks Underlying Human Diseases, Their Comorbidities and Their Druggable Targets.

The genetic basis of complex diseases involves alterations on multiple genes. Unraveling the interplay between these genetic factors is key to the discovery of new biomarkers and treatments. In 2014, we introduced GUILDify, a web server that searches for genes associated to diseases, finds novel disease genes applying various network-based prioritization algorithms and proposes candidate drugs. Here, we present GUILDify v2.0, a major update and improvement of the original method, where we have included protein interaction data for seven species and 22 human tissues and incorporated the disease-gene associations from DisGeNET. To infer potential disease relationships associated with multi-morbidities, we introduced a novel feature for estimating the genetic and functional overlap of two diseases using the top-ranking genes and the associated enrichment of biological functions and pathways (as defined by GO and Reactome). The analysis of this overlap helps to identify the mechanistic role of genes and protein-protein interactions in comorbidities. Finally, we provided an R package, guildifyR, to facilitate programmatic access to GUILDify v2.0 (http://sbi.upf.edu/guildify2).

Pancreatic cancer and autoimmune diseases: An association sustained by computational and epidemiological case-control approaches.

Deciphering the underlying genetic basis behind pancreatic cancer (PC) and its associated multimorbidities will enhance our knowledge toward PC control. The study investigated the common genetic background of PC and different morbidities through a computational approach and further evaluated the less explored association between PC and autoimmune diseases (AIDs) through an epidemiological analysis. Gene-disease associations (GDAs) of 26 morbidities of interest and PC were obtained using the DisGeNET public discovery platform. The association between AIDs and PC pointed by the computational analysis was confirmed through multivariable logistic regression models in the PanGen European case-control study population of 1,705 PC cases and 1,084 controls. Fifteen morbidities shared at least one gene with PC in the DisGeNET database. Based on common genes, several AIDs were genetically associated with PC pointing to a potential link between them. An epidemiologic analysis confirmed that having any of the nine AIDs studied was significantly associated with a reduced risk of PC (Odds Ratio (OR) = 0.74, 95% confidence interval (CI) 0.58-0.93) which decreased in subjects having ≥2 AIDs (OR = 0.39, 95%CI 0.21-0.73). In independent analyses, polymyalgia rheumatica, and rheumatoid arthritis were significantly associated with low PC risk (OR = 0.40, 95%CI 0.19-0.89, and OR = 0.73, 95%CI 0.53-1.00, respectively). Several inflammatory-related morbidities shared a common genetic component with PC based on public databases. These molecular links could shed light into the molecular mechanisms underlying PC development and simultaneously generate novel hypotheses. In our study, we report sound findings pointing to an association between AIDs and a reduced risk of PC.

Embracing the Dark Side: Computational Approaches to Unveil the Functionality of Genes Lacking Biological Annotation in Drug-Induced Liver Injury.

In toxicogenomics, functional annotation is an important step to gain additional insights into genes with aberrant expression that drive pathophysiological mechanisms. Nevertheless, there exists a gap on annotation of these genes which often hampers the interpretation of results and limits their applicability in translational medicine. In this study, we evaluated the coverage of functional annotations of differentially expressed genes (DEGs) induced by 10 selected compounds from the TG-GATEs database identified as high- or no-risk in causing drug-induced liver injury (most-DILI or no-DILI, respectively) using in vitro human data. Functional roles of DEGs not present in the most common biological annotation databases - termed "dark genes" - were unveiled via literature mining and via the identification of shared regulatory transcription factors or signaling pathways. Our results demonstrated that there were approximately 13% of dark genes induced by these compounds in vitro and we were able to obtain additional relevant information for up to 76% of those. Using interactome data from several sources, we have uncovered genes such as LRBA, and WDR26 as highly connected in the protein network that play roles in drug response. Genes such as MALAT1, H19, and MIR29C - whose links to hepatotoxicity have been confirmed - were identified as markers for the most-DILI group and appeared as top hits across all literature-based mining methods. Furthermore, we investigated the potential impact of dark genes on liver toxicity by identifying their rat orthologs in combination with their correlation to drug-induced liver pathologies observed in vivo following chemical exposure. We identified a set of important regulatory transcription factors of dark genes for all most-DILI compounds including E2F1 and JUND with supporting evidences in literature and we found Magee1 correlated with chemically induced bile duct hyperplasia and adverse responses at 29 days in rats in vivo. In conclusion, in this study we show the potential role of these poorly annotated genes in mechanisms underlying hepatotoxicity and offer a number of computational approaches that may help to minimize current gaps in gene annotation and highlight their values as potential biomarkers in toxicological studies.