CoMent: Relationships Between Biomedical Concepts Inferred From the Scientific Literature

J Mol Biol. 2022 Jun 15;434(11):167568. doi: 10.1016/j.jmb.2022.167568. Epub 2022 Mar 30.


The mining of the massive amounts of biomedical information is hindered by the still scarce representation of these data using formal vocabularies and ontologies, which is necessary for cross-linking conceptual entities between different resources and, in general, representing the information in a computer-tractable way. Basic things such as retrieving a comprehensive list of associations between complex diseases and their reported symptoms or underlying biological processes, given in terms of formal identifiers, are not trivial and, in many cases, these have to be generated by manual curation or inferred/predicted from indirect evidences. In this work, using a text-mining approach based on detecting significant co-mentions in the scientific literature, we generated a resource with millions of relationships between thousands of terms representing diseases, symptoms, biological processes, molecular functions and cellular compartments, all given in terms of formal identifiers of these terms in the main resources dealing with them. We show some examples that highlight the differences between these relationships and those that are available in other resources. These relationships can be queried and inspected in an interactive web interface freely available at:

Keywords: biological process; biomedical ontology; disease; molecular function; symptom.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Computational Biology*
  • Data Mining*