Integrated query of three open-source bioinformatics tools to reveal clinical and molecular insights into environmental determinants of disease

BACKGROUND: We have developed a suite of open-source clinical and biomedical informatics applications to facilitate translational science. The Integrated Clinical and Environmental Exposures Service (ICEES) provides regulatory-compliant access to integrated clinical and environmental exposures data. Reasoning Over Biomedical Objects linked in Knowledge Oriented Pathways (ROBOKOP) is a knowledge graph (KG)-based system for exploration of integrated and harmonized biomedical knowledge sources. Causal Activity Model (CAM) Knowledge Provider (KP) exposes causal inferences between biomedical entities and the connections between them. The applications are interoperable and can be queried via a common endpoint or independently to yield unique clinical and biomedical insights. In this study, we aimed to sequentially query the three independent applications to demonstrate their complementary nature and the value of a user-driven approach to derive mechanistic insights into environmental determinants of asthma, our application use case. METHODS: We developed a use case on environmental determinants of asthma, focusing specifically on exposure to airborne particulate matter and its effects on asthma exacerbations. We framed our queries in the context of the proposed role of toll-like receptors (TLRs) in asthma. We first queried ICEES via an open application programming interface to examine the relationship between exposure to airborne particulate matter and asthma exacerbations among patients with chronic pulmonary disease. We then used the ICEES results as input for query of the ROBOKOP question-builder to explore biological mechanisms that might explain the ICEES-derived clinical observations on particulate matter exposure and asthma by way of establishing connections in the ROBOKOP KG that link particulate matter to asthma. The CAM KP user interface was queried to extend the mechanistic insights derived from ROBOKOP to identify causal relationships between the biomedical entities. RESULTS: ICEES showed that exposure to high levels of airborne particulate matter was associated with an increase in asthma exacerbations, with African Americans showing higher overall rates of asthma exacerbations than Caucasians. ROBOKOP established connections between particulate matter and asthma by first establishing that a particulate matter - TLR4 - MYD88 connection exists in the ROBOKOP KG, as suggested in the literature, and then exploring genes and biological processes or activities that connect MYD88 to asthma. We highlight an example assertion that: particulate matter increases the response to TLR4, which physically and genetically interacts with MYD88, which catalyzes cytokine receptor binding by IL-5, which is genetically associated with asthma. CAM KP extended the ROBOKOP findings to identify a causal relationship between the TLR4 pathway and the immunologically relevant genes LY96, CD14, RELA, and TNF. CONCLUSIONS: We conclude that user-driven sequential query of ICEES, ROBOKOP, and CAM KP provides a more complete understanding of the role of environmental determinants in disease than each service can provide independently. The three open applications are interoperable and complementary and together provide powerful tools to derive mechanistic insights into asthma and other diseases.