Graph deep learning reveals multiple signal pathways activated in anti-citrullinated protein antibodies stimulated synoviocytes

Abstract

AO_SCPLOWBSTRACTC_SCPLOWWe present a study of anti-citrullinated protein antibodies stimulated fibroblast-like synoviocytes from patients with rheumatoid arthritis, which is powered by a novel graph deep learning framework applied to single-cell mRNA expression data. This new analytical framework discovered dominant pathways that suggest IL1-IL1R mediated signaling, a novel signal transduction response to ACPA-stimulation, and comfirmed our previous finding of PI3K/AKT activation in response of ACPA-stimulation as well. The study demonstrated the capability of graph deep learning framework on signaling pathway analysis. The findings suggest that ACPAs contribute to distinct pathogenic events through activating multiple signaling pathways.