The N3C governance ecosystem: A model socio-technical partnership for the future of collaborative analytics at scale

The National COVID Cohort Collaborative (N3C) is a public-private-government partnership established during the Coronavirus pandemic to create a centralized data resource called the "N3C data enclave." This resource contains individual-level health data from participating healthcare sites nationwide to support rapid collaborative analytics. N3C has enabled analytics within a cloud-based enclave of data from electronic health records from over 17 million people (with and without COVID-19) in the USA. To achieve this goal of a shared data resource, N3C implemented a shared governance strategy involving stakeholders in decision-making. The approach leveraged best practices in data stewardship and team science to rapidly enable COVID-19-related research at scale while respecting the privacy of data subjects and participating institutions. N3C balanced equitable access to data, team-based scientific productivity, and individual professional recognition - a key incentive for academic researchers. This governance approach makes N3C research sustainable and effective beyond the initial days of the pandemic. N3C demonstrated that shared governance can overcome traditional barriers to data sharing without compromising data security and trust. The governance innovations described herein are a helpful framework for other privacy-preserving data infrastructure programs and provide a working model for effective team science beyond COVID-19.

Risk of post-acute sequelae of SARS-CoV-2 infection associated with pre-coronavirus disease obstructive sleep apnea diagnoses: an electronic health record-based analysis from the RECOVER initiative

STUDY OBJECTIVES

Obstructive sleep apnea (OSA) has been associated with more severe acute coronavirus disease-2019 (COVID-19) outcomes. We assessed OSA as a potential risk factor for Post-Acute Sequelae of SARS-CoV-2 (PASC).

METHODS

We assessed the impact of preexisting OSA on the risk for probable PASC in adults and children using electronic health record data from multiple research networks. Three research networks within the REsearching COVID to Enhance Recovery initiative (PCORnet Adult, PCORnet Pediatric, and the National COVID Cohort Collaborative [N3C]) employed a harmonized analytic approach to examine the risk of probable PASC in COVID-19-positive patients with and without a diagnosis of OSA prior to pandemic onset. Unadjusted odds ratios (ORs) were calculated as well as ORs adjusted for age group, sex, race/ethnicity, hospitalization status, obesity, and preexisting comorbidities.

RESULTS

Across networks, the unadjusted OR for probable PASC associated with a preexisting OSA diagnosis in adults and children ranged from 1.41 to 3.93. Adjusted analyses found an attenuated association that remained significant among adults only. Multiple sensitivity analyses with expanded inclusion criteria and covariates yielded results consistent with the primary analysis.

CONCLUSIONS

Adults with preexisting OSA were found to have significantly elevated odds of probable PASC. This finding was consistent across data sources, approaches for identifying COVID-19-positive patients, and definitions of PASC. Patients with OSA may be at elevated risk for PASC after SARS-CoV-2 infection and should be monitored for post-acute sequelae.

Biolink Model: A universal schema for knowledge graphs in clinical, biomedical, and translational science

Within clinical, biomedical, and translational science, an increasing number of projects are adopting graphs for knowledge representation. Graph‐based data models elucidate the interconnectedness among core biomedical concepts, enable data structures to be easily updated, and support intuitive queries, visualizations, and inference algorithms. However, knowledge discovery across these “knowledge graphs” (KGs) has remained difficult. Data set heterogeneity and complexity; the proliferation of ad hoc data formats; poor compliance with guidelines on findability, accessibility, interoperability, and reusability; and, in particular, the lack of a universally accepted, open‐access model for standardization across biomedical KGs has left the task of reconciling data sources to downstream consumers. Biolink Model is an open‐source data model that can be used to formalize the relationships between data structures in translational science. It incorporates object‐oriented classification and graph‐oriented features. The core of the model is a set of hierarchical, interconnected classes (or categories) and relationships between them (or predicates) representing biomedical entities such as gene, disease, chemical, anatomic structure, and phenotype. The model provides class and edge attributes and associations that guide how entities should relate to one another. Here, we highlight the need for a standardized data model for KGs, describe Biolink Model, and compare it with other models. We demonstrate the utility of Biolink Model in various initiatives, including the Biomedical Data Translator Consortium and the Monarch Initiative, and show how it has supported easier integration and interoperability of biomedical KGs, bringing together knowledge from multiple sources and helping to realize the goals of translational science.

The landscape of nutri-informatics: a review of current resources and challenges for integrative nutrition research

Informatics has become an essential component of research in the past few decades, capitalizing on the efficiency and power of computation to improve the knowledge gained from increasing quantities and types of data. While other fields of research such as genomics are well represented in informatics resources, nutrition remains underrepresented. Nutrition is one of the most integral components of human life, and it impacts individuals far beyond just nutrient provisions. For example, nutrition plays a role in cultural practices, interpersonal relationships and body image. Despite this, integrated computational investigations have been limited due to challenges within nutrition informatics (nutri-informatics) and nutrition data. The purpose of this review is to describe the landscape of nutri-informatics resources available for use in computational nutrition research and clinical utilization. In particular, we will focus on the application of biomedical ontologies and their potential to improve the standardization and interoperability of nutrition terminologies and relationships between nutrition and other biomedical disciplines such as disease and phenomics. Additionally, we will highlight challenges currently faced by the nutri-informatics community including experimental design, data aggregation and the roles scientific journals and primary nutrition researchers play in facilitating data reuse and successful computational research. Finally, we will conclude with a call to action to create and follow community standards regarding standardization of language, documentation specifications and requirements for data reuse. With the continued movement toward community standards of this kind, the entire nutrition research community can transition toward greater usage of Findability, Accessibility, Interoperability and Reusability principles and in turn more transparent science.

A harmonized meta-knowledgebase of clinical interpretations of somatic genomic variants in cancer

Precision oncology relies on accurate discovery and interpretation of genomic variants, enabling individualized diagnosis, prognosis and therapy selection. We found that six prominent somatic cancer variant knowledgebases were highly disparate in content, structure and supporting primary literature, impeding consensus when evaluating variants and their relevance in a clinical setting. We developed a framework for harmonizing variant interpretations to produce a meta-knowledgebase of 12,856 aggregate interpretations. We demonstrated large gains in overlap between resources across variants, diseases and drugs as a result of this harmonization. We subsequently demonstrated improved matching between a patient cohort and harmonized interpretations of potential clinical significance, observing an increase from an average of 33% per individual knowledgebase to 57% in aggregate. Our analyses illuminate the need for open, interoperable sharing of variant interpretation data. We also provide a freely available web interface (search.cancervariants.org) for exploring the harmonized interpretations from these six knowledgebases.

How many rare diseases are there?

A lack of robust knowledge of the number of rare diseases and the number of people affected by them limits the development of approaches to ameliorate the substantial cumulative burden of rare diseases. Here, we call for coordinated efforts to more precisely define rare diseases.

The Human Phenotype Ontology in 2017

Deep phenotyping has been defined as the precise and comprehensive analysis of phenotypic abnormalities in which the individual components of the phenotype are observed and described. The three components of the Human Phenotype Ontology (HPO; www.human-phenotype-ontology.org) project are the phenotype vocabulary, disease-phenotype annotations and the algorithms that operate on these. These components are being used for computational deep phenotyping and precision medicine as well as integration of clinical data into translational research. The HPO is being increasingly adopted as a standard for phenotypic abnormalities by diverse groups such as international rare disease organizations, registries, clinical labs, biomedical resources, and clinical software tools and will thereby contribute toward nascent efforts at global data exchange for identifying disease etiologies. This update article reviews the progress of the HPO project since the debut Nucleic Acids Research database article in 2014, including specific areas of expansion such as common (complex) disease, new algorithms for phenotype driven genomic discovery and diagnostics, integration of cross-species mapping efforts with the Mammalian Phenotype Ontology, an improved quality control pipeline, and the addition of patient-friendly terminology.

Expression Atlas update--a database of gene and transcript expression from microarray- and sequencing-based functional genomics experiments

Expression Atlas (http://www.ebi.ac.uk/gxa) is a value-added database providing information about gene, protein and splice variant expression in different cell types, organism parts, developmental stages, diseases and other biological and experimental conditions. The database consists of selected high-quality microarray and RNA-sequencing experiments from ArrayExpress that have been manually curated, annotated with Experimental Factor Ontology terms and processed using standardized microarray and RNA-sequencing analysis methods. The new version of Expression Atlas introduces the concept of 'baseline' expression, i.e. gene and splice variant abundance levels in healthy or untreated conditions, such as tissues or cell types. Differential gene expression data benefit from an in-depth curation of experimental intent, resulting in biologically meaningful 'contrasts', i.e. instances of differential pairwise comparisons between two sets of biological replicates. Other novel aspects of Expression Atlas are its strict quality control of raw experimental data, up-to-date RNA-sequencing analysis methods, expression data at the level of gene sets, as well as genes and a more powerful search interface designed to maximize the biological value provided to the user.

Prediction of immunogenicity: in silico paradigms, ex vivo and in vivo correlates

Immunogenicity can be a major obstacle to successful protein drug therapy. Antidrug antibodies may neutralize therapeutic function, influence pharmacokinetics and, in some cases, lead to severe adverse effects. These effects depend on factors including dose, regimen, delivery route and contaminants, among others. Importantly, immunogenicity is a consideration that is better addressed during preclinical development before complications arise in clinical trials or following licensure. This article will address the development and application of computational tools for immunogenicity assessment of protein therapeutics, and validation of those predictions using peripheral blood from exposed subjects or alternative in vivo methods.

Novel function of complement C3d as an autologous helper T-cell target

The C3d fragment of complement component C3 has been shown to enhance immune responses to antigens that lack T-cell epitopes such as bacterial polysaccharides. C3d binds to the B-cell complement receptor 2 (CR2 or CD21); this binding serves as a co-activation signal to the B cell when the polysaccharide antigen portion binds simultaneously to the B-cell receptor (surface Ig). Bringing together receptor-associated signal transduction molecules CD19 and Igalpha/beta, respectively, results in a lower threshold of activation. Paradoxically, C3d has also been shown to enhance antibody titers in the CD21 knockout (KO) mouse model as well as increase Th1 and Th2 cytokine secretion, suggesting that that an auxiliary CR2-independent pathway of immune activation may exist. We hypothesized that in addition to its molecular adjuvant property that enhances signal 1 during B-cell activation (co-signal 1), C3d also contains T-cell epitopes that are able to stimulate autoreactive C3d peptide-specific helper T cells which we term 'co-signal 2'. Using the EpiMatrix T-cell epitope-mapping algorithm, we identified 11 putative T-cell epitopes in C3d, a very high epitope density for a 302 amino-acid sequence. Eight of these epitope candidates were synthesized and shown to bind a variety of class II HLA-DR molecules of different haplotypes, and to stimulate C3d peptide-specific T cells to secrete pro-inflammatory cytokines in vitro. Further, we demonstrate a C3d-peptide specific increase in CD4(+) intracellular IFN-gamma(+) T cells in peripheral blood mononuclear cells (PBMCs) exposed to C3d peptides in vitro. We believe that the discovery of these autologous T cells autoreactive for C3d provides evidence supporting the 'co-signal 2' hypothesis and may offer a novel explanation of the CD21 KO paradox.

Developing an epitope-driven tuberculosis (TB) vaccine

Epitope-driven vaccines are created from selected sub-sequences of proteins, or epitopes, derived by scanning the protein sequences of pathogens for patterns of amino acids that permit binding to human MHC molecules. We developed a prototype tuberculosis (TB) vaccine that contains epitopes derived by (1) EpiMer mapping of previously published secreted proteins derived from Mycobacterium tuberculosis (Mtb), and (2) EpiMatrix mapping of selected Mtb genome open reading frames (ORFs). Each of the epitopes contains at least three distinct class II MHC binding motif matches. These Mtb epitope selections were validated by measuring T cell responses from peripheral blood mononuclear cells (PBMC) obtained from healthy, asymptomatic tuberculin skin test-positive donors. Twenty-four validated Mtb epitopes were selected for inclusion in a DNA plasmid vector. We immunized HLA-DR B*0101 transgenic mice with this vaccine prototype augmented by co-administration of rIL-15. Following administration of three immunizations at 14-day intervals in conjunction with rIL-15, epitope-specific T cell responses were observed to eight of the 24 epitopes contained in the DNA construct, one week following the last injection. The systematic application of bioinformatics tools to whole genomes, in combination with in vitro methods for screening and confirming epitopes, may lead to the development of novel vaccines for infectious diseases like TB.

Analyzing Mycobacterium tuberculosis proteomes for candidate vaccine epitopes

Secreted antigens of Mycobacterium tuberculosis (Mtb) induce strong T cell responses and interferon-gamma (IFN-gamma) secretion, both of which are integral in the defense against Mtb. We used web-based tools (SignaIP and Prosite) to identify putative secreted proteins from Mtb genomes CDC 1551 and H37Rv. We then used EpiMatrix, a proprietary pattern-matching algorithm, to do a preliminary analysis of these proteins for regions that contained a high number of class II MHC binding motif matches. The use of bioinformatics tools reduced the number of potential epitopes to be screened to 5% of the 1.3 million overlapping peptides. Peripheral blood mononuclear cells (PBMC) were obtained from healthy, asymptomatic tuberculin skin test-positive donors. Of the 17 highest-ranking peptide candidates that could be synthesized for this preliminary in vitro evaluation, 15 (88%) stimulated IFN-gamma response, and eight (47%) stimulated lymphocyte proliferation in vitro. IFN-gamma ELISpot assays were therefore a more sensitive test for T cell response to these peptides than were proliferation assays. One highly promiscuous epitope (MT2281-26-J, WRRRPLSSALLSFGLLLGGLPL) induced IFN-gamma secretion in PBMC from 11 of 25 Mtb immune subjects (44%). Overall, 15 epitopes, and MT2281-26-J in particular, are candidates for inclusion in a multi-epitope TB vaccine. These findings support the systematic application of bioinformatics tools to whole genomes when used in combination with in vitro methods for screening and confirming epitopes.

Immuno-informatics: Mining genomes for vaccine components

The complete genome sequences of more than 60 microbes have been completed in the past decade. Concurrently, a series of new informatics tools, designed to harness this new wealth of information, have been developed. Some of these new tools allow researchers to select regions of microbial genomes that trigger immune responses. These regions, termed epitopes, are ideal components of vaccines. When the new tools are used to search for epitopes, this search is usually coupled with in vitro screening methods; an approach that has been termed computational immunology or immuno-informatics. Researchers are now implementing these combined methods to scan genomic sequences for vaccine components. They are thereby expanding the number of different proteins that can be screened for vaccine development, while narrowing this search to those regions of the proteins that are extremely likely to induce an immune response. As the tools improve, it may soon be feasible to skip over many of the in vitro screening steps, moving directly from genome sequence to vaccine design. The present article reviews the work of several groups engaged in the development of immuno-informatics tools and illustrates the application of these tools to the process of vaccine discovery.