About This Project

Using our recently developed COVID-19 Infectious Disease Ontology in combination with the existing Coronavirus Infectious Disease Ontology, we will annotate 400 articles in PubMed on coronavirus research. The resulting gold standard will be used to train algorithms for automated annotation tasks, which will be used to identify patterns in COVID-19 datasets, with applications including data mining and consolidation, pattern recognition, and hypothesis generation and testing.

Ask the Scientists

What is the context of this research?

Researchers have rapidly accumulated data relevant to SARS-CoV-2 and - COVID-19 - from a variety of sources in efforts to combat the virus (see examples in The Pivotal Role of TMPRSS2, Sex Differences in Mortality from COVID-19, Gender Differences in Patients with COVID-19). There is too much data generated to examine manually, and too little time to waste. Pathogenic patterns will invariably be overlooked by even the most attentive researchers. One solution is to use computational support and ontologies to explore patterns across datasets. Ontologies are structured vocabularies, already widely used in bioinformatics (as discussed here, here, and here) and biomedical data standardization, which support data integration, sharing, reproducibility, and automated reasoning.

What is the significance of this project?

Our COVID-19 Infectious Disease Ontology (IDO-COVID-19) - ontology details discussed here; ontologies found on Bioportal here) will enable rapid integration, discoverability, and reproducibility, of COVID-19 data across disparate medical research datasets, and provide needed data infrastructure to support machine learning and automated reasoning over such datasets. This will ensure the data is more easily harvestable for use in public health efforts to combat the COVID-19 pandemic. In that respect, our project has the potential to impact the COVID-19 pandemic similarly to the way the Gene Ontology has been used, with enormous success [https://doi.org/10.1093/nar/gky1055], to facilitate data comparability in genome-related science.

What are the goals of the project?

Annotation of heterogeneous datasets using common ontologies links their data together by means of the semantically controlled properties of ontology terms, thus facilitating data integration. IDO-COVID-19 will be used to annotate approximately 400 articles in the National Library of Medicine COVID-19 corpus, which reports COVID-19 clinical trial, epidemiological, and pathogenesis data. This will provide a natural language processing ‘gold standard’ used to train algorithms for automated annotation tasks, which will, in turn, be used to identify useful patterns in COVID-19 datasets, with varying applications including data mining and consolidation, and pattern recognition.