Fever, or pyrexia, is an abnormal elevation of body temperature, usually a result of a pathologic process. Normal body temperature is ranged 98-100°F (36.5-37.5°C), and temperatures above this range are usually considered febrile. Increased body temperature usually indicates possible presence of infection or sepsis. Once an infection occurs, the body responds to control the infection, often resulting in increased temperature. The fever in response to infection is likely a cure to remove infection and create a favorable environment for immune compartments such as white blood cells [1
]. Nevertheless, a long-lasting fever can cause devastating effects; therefore, reducing fever either with medication or physical cooling methods remains a common practice.
Fever is induced by a substance called pyrogen, which can be either endogenous or exogenous to the body. Endogenous pyrogens are cytokines produced by phagocytic cells. Major endogenous pyrogens include interleukin-1 α/β (IL-1A/B), interleukin 6 (IL6), and tumor necrosis factor alpha (TNFA) [2
]. Minor pyrogens include interleukin-8 (IL8) and interferon-α/β/γ (INF-A/B/G). These major or minor endogenous pyrogens are released into the general circulation, migrate to the circumventricular organs of the brain, and activate the arachidonic acid pathway. Exogenous pyrogens, such as lipopolysaccharide (LPS) from Gram-negative bacteria, can interact with host cell immune factors, such as LPS-binding protein (LBP), and trigger the release of endogenous factors, which in turn activate the arachidonic acid pathway [2
]. The arachidonic acid pathway is mediated by phospholipase A2 (PLA2), cyclooxygenase-2 (COX-2), and prostaglandin E2 synthases (PTGES) [3
]. These enzymes mediate the synthesis and release of prostaglandin E2 (PGE2). PGE2, the ultimate mediator of the fever response, stimulates the hypothalamus in the brain to generate a systemic response to increase the body temperature. The hypothalamus is responsible for coordinating complex heat effector mechanisms [4
]. While the general fever pathway has been well studied, more detailed gene interaction networks associated with fever under different experimental conditions are typically unclear.
Vaccination is the process of administration of a vaccine to a host to stimulate the host immune system to develop adaptive immunity to a pathogen or against a specific disease (e.g., cancer). The immunological process after vaccination involves many immune cells including macrophages, dendritic cells, and lymphocytes. These immune cells can undergo certain levels of inflammation enhanced by various immune factors. Many vaccines can frequently cause fever [5
]. Our main hypothesis is that vaccination stimulates inflammatory fever responses that may be required for the induction of protective immunity or act as an undesired adverse effect. However, how vaccination perturbs certain fever-related genes to cause the adverse event is still unclear. This study targets learning more about the genetic interaction processes behind the vaccine-induced fever inflammatory responses.
We previously demonstrated that high-throughput literature mining and the use of ontology can significantly enhance our understanding of vaccine research [9
]. First, we developed a literature-based discovery (LBD) approach integrating text mining with network centrality analysis, which was successfully applied to a study investigating the role of interferon-gamma (IFN-γ) in vaccine-mediated gene-interaction networks [9
]. Here an interaction network represents a network with various direct and indirect interactions. Gene-gene interaction networks were generated from the biomedical literature using natural language processing (NLP) techniques, and the most important genes in these networks were identified by network centrality analyses using four types of centrality measures: degree, eigenvector, closeness, and betweenness. Integrating these multiple centrality-based core gene sets in the vaccine subdomain resulted in the identification of a vaccine-specific sub-network of IFN-γ [9
In an extended study, the application of the Vaccine Ontology (VO) significantly improved the analysis of the vaccine-specific IFN-γ sub-network [11
]. A biomedical ontology is a controlled set of terms and relations that represent entities in the scientific world (e.g., the vaccine domain) and how they relate to each other. Therefore, a biomedical ontology can be considered as a well-defined machine-parsable “terminology” of terms together with logically defined relations between these terms. VO is a community-based ontology in the domain of vaccine and vaccination [12
]. Developed in Web Ontology Language (OWL), VO provides a logic based framework for describing associations of vaccines (including licensed vaccines, vaccines in clinical trial, or vaccines proven in research), vaccine components, microbial genes engineered for vaccine development, and vaccine-induced host gene and immune responses. The relations between different VO vaccine terms have been logically defined and support advanced semantic reasoning.
Specifically, VO can be useful in two ways in the literature mining-based approach for vaccine research [11
]. First, VO provides an asserted list of specific vaccines and synonyms of each vaccine, allowing the extraction of interactions between IFN-γ and specific vaccines (instead of the general term “vaccine” from sentences). Secondly, the rich semantic constructs in the VO OWL format (e.g.
, necessary and sufficient conditions) provides logical definitions (axioms) of vaccine attributes and enables the inference of the subclasses of additional parent terms (e.g.
, “inactivated vaccines”). VO includes a hierarchical structure based on transitive “is_a” relation. This relation indicates that a child term (e.g.
, “M. tuberculosis
vaccine BCG”) is always a parent term (e.g.
, “M. tuberculosis
vaccine”) with a specific restriction. The attributes of a specific vaccine are also defined in VO. For example, BCG is defined to have the quality of “virulent” and “viable (synonym: live)”. Necessary and sufficient conditions can also be used for inference. For instance, although BCG is not asserted as a child of “live attenuated vaccine”, the BCG “is_a” hierarchy definition combined with its attributes will allow a reasoner to infer BCG as a “live attenuated vaccine”.
Using VO, we obtained more genes and gene interactions from the vaccine-mediated IFN-γ-gene interaction network, and were also able to classify identified genes and gene interactions using the asserted and inferred hierarchies of different vaccines [11
]. Another study from our group demonstrated that VO-based literature mining provided a better performance in retrieving Brucella
vaccine-related literature and building gene interaction networks than the Medical Subject Headings (MeSH)-based approach [10
]. These studies helped to generate new candidate genes for vaccine development.
The general literature mining strategy that integrates centrality and ontology has been named by us as the CONDL, standing for Centrality and Ontology-based Network Discovery using Literature data [11
]. Here we report the application of the CONDL approach to retrieve gene-gene and gene-vaccine interaction networks associated with fever or vaccine-associated fever processes. Central genes and enriched biological functions are identified in these interaction networks.