Psoriatic arthritis disease subtypes are phenocopied in a novel humanized murine model of psoriasis and joint inflammation

Data processing and analysis

Data normalization, quality control, and analysis was conducted by the UPMC Hillman Cancer Bioinformatics Services. Following the GeoMX NGS pipeline (sequencing saturation 92%±0.12% (mean±S.E.M)), the resulting read count matrix was passed through robust QC procedure to remove ROIs of low surface area (<5000 µm2) or low perfect aligned reads (<80%), and probes of low quality or identified as global or local outliers (fails Grubbs outlier test in ≥ 20% segments). The remaining data were back-ground subtracted and Q3 normalized to house-keeping genes. ROIs were considered positive for CD8 T cells if their CD8 memory T cell deconvolution fraction was > 0 (R, spatialDecon, CellProfileLibrary; Human Adult ImmuneTumor_safeTME profile). Genes differentially expressed between groups of interest were detected using gene-wise linear modeling in R (edgeR, limma v3.50.3), with p-value adjusted for false discovery rate. Contrasts of groups were performed using t-tests at a significance level of 0.05.