Skip to main content
Fig. 4 | Arthritis Research & Therapy

Fig. 4

From: Network analysis of synovial RNA sequencing identifies gene-gene interactions predictive of response in rheumatoid arthritis

Fig. 4

Gene pair interactions linked to endothelial activation and Akt signaling enhance prediction of response to methotrexate-based therapy. A Network of unique active interactions in conventional synthetic DMARD poor responders. Cluster NR1. Antigen processing and presentation cluster (100% enrichment). Cluster NR2. Genes of the chemokine signaling pathway (100% enrichment). Cluster NR3. Cluster associated to Wnt signaling pathway (74.2% enrichment). Cluster NR4. Cluster linked to VEGFR2 mediated vascular permeability (38.8% enrichment). Red boxes highlight predictive gene pairs. B–E, H, L Robust linear regression with interaction term associated with response for B GNAI3 and CXCR5 C NOS3 and CAMK1, D NOS3 and AKT3 E AKT1 and PPP2R3B, H NOS3 and CAMK2D, L ATP1B1 and PIK3CD. p-values of the interacting terms are shown. F, I, M Logistic regression of response as a function of F AKT1 and PPP2R3B, I NOS3 and CAMK2D, M ATP1B1 and PIK3CD. p-values of the response to gene interacting term are shown. Expression of the second gene is dichotomized at ± 1 standard deviation. G, J, N Receiver operating characteristic (ROC) curve analysis of the of robust linear model ability to predict response using G AKT1 and PPP2R3B J NOS3 and CAMK2D N ATP1B1 and PIK3CD. All plots show a ROC curve for both the model including the gene-gene ratio interaction term (in black) and the equivalent model excluding the ratio

Back to article page