Biological links in periodontitis and rheumatoid arthritis: Discovery via text-mining PubMed abstracts.

BACKGROUND AND OBJECTIVE: Primary research concerning molecular pathways that link rheumatoid arthritis with periodontitis is limited. Biomedical literature data mining can offer insights into putative linkage mechanisms toward hypothesis development, based on information discovery. The aim of this study was to explore potential Periodontitis-Rheumatoid Arthritis biological links by analysing "overlapping" genes reported in biomedical abstracts. MATERIAL AND METHODS: PubMed abstracts for terms: (a) "Periodontitis" or "Periodontal Diseases" (PD), (b) "Rheumatoid arthritis" (RA), and (c) their combination with "AND" (RA+PD), were each text-mined to extract genes using "Human Genome Nomenclature Committee" (HGNC) symbols. A gene-set common to RA and PD abstracts was determined (RA intersectionPD). Gene ontology (GO) profiles of RA intersectionPD and RA+PD were compared using "GoProfiler." Minimum order protein-protein interaction (PPI) and gene-miRNA networks of "differential genes" between RA intersectionPD and RA+PD were constructed with "networkAnalyst." RESULTS: Among 1676 genes documented in RA (10 5241 abstracts), and 893 genes in PD (80 982 abstracts), 535 genes were common (RA intersectionPD), from which 35 genes were also documented in RA+PD (415 abstracts). 41 GO-terms significantly different between RA intersectionPD and RA+PD GO profiles represented 38 biological processes including; nitric oxide metabolism, immunoglobulin production, hormonal regulation, catabolic process down-regulation, and leukocyte proliferation. The 500 differential genes' PPI and gene-miRNA networks showed REL, TRAF2, AQP1 genes, and miRNAs 335-5p, 17-5p, 93-5p with genes HMOX1 and SP1 as hub nodes. CONCLUSIONS: Text-mining biomedical abstracts revealed potentially shared but un-investigated links between PD and RA, meriting further research.