An expanding literature reveals that gut microbiota produce and modulate the production of neurotransmitters. These neurotransmitters, like serotonin, dopamine, GABA, norepinephrine, histamine, and acetylcholine, are known to be associated with distress and mental health impairment. In an attempt to gather these rather dispersed results and effectively provide an overview of knowledge in this area, researchers have recently developed a “gut microbiota knowledge graph for mental disorders” called MiKG4MD (Liu et al.).
The researchers’ methodology involved collecting data sources and performing data extraction and structure, knowledge base enrichment, and knowledge graph visualization (Liu et al.). To begin, researchers looked for Google Scholar and PubMed literature to identify studies examining the link between gut microbiota and neurotransmitters. Thirty-five articles were identified, and the evidence levels of these studies were ranked from A to E in accordance with the strength of trial design. Several classes of annotations were created for this particular study and were divided among “entities” such as neurotransmitter, gut microbiota, and mental disorder (Liu et al.). As the extraction of entities and semantic relations is considered vital to construct a knowledge graph, researchers underwent this process manually. The Terse RDF Triple Language (Turtle) format was used to structure these entities and concepts (Liu et al.).
Next, existing biomedical ontologies/terminologies were used to enrich the semantic database (Liu et al.). This included databases like the UMLS which covers medical terminologies and KEGG which provides information on chemical substances and pathway maps of molecular interaction.
Finally, three test cases were designed to demonstrate the performance of the knowledge graph in terms of identification and prediction. To use the knowledge graph, one has to use a SPARQL query, which has three components: the PREFIX defining the ontologies to be used in the query, the SELECT DISTINCT statement used to return only distinct values, and the WHERE clause used to specify certain conditions for the records one wants to extract (Liu et al.). Researchers used a gut microbiota-based query, a neurotransmitter-based query, and a mental disorder-based query. The first case involved questioning if Bifidobacterium dentium may be linked to anxiety and depressive disorders via neurotransmitter regulation. Researchers queried the variables related to the microbiota, including neurotransmitters, reference, and mental disorder. The results stated that GABA is the only neurotransmitter regulated by Bifidobacterium dentium, and that it may be related to anxiety and depressive disorders via this regulatory relationship. The remaining two test cases yielded similar results (Liu et al.).
The MiKG4MD seems incredibly promising in how it provides a means to both achieve semantic queries and support medical researchers in making decisions regarding therapy implementation for mental disorders. Though MiKG4MD currently does have limitations—particularly in terms of its limited coverage of the knowledge domain and the fact that it only yields one bacterium-one neurotransmitter results (which is all too simplistic)—the possibility of further automated enrichment of the knowledge graph is exciting.
Liu, Ting, et al. "Predicting the Relationships between Gut Microbiota and Mental Disorders with Knowledge Graphs." Health Information Science and Systems, vol. 9, no. 1, 2021, https://doi.org/10.1007/s13755-020-00128-2.