Fig. 1

Computational pipeline to predict psoriasis subtypes. a Overview of pipeline, through quality control, phasing and imputation, association analysis, meta-analysis, and stepwise conditional analysis. b The machine-learning process included separating data randomly into training (cross-validation to optimize the model) and test (holdout) sets, as well as evaluating the results with and without the PAGE Immunochip dataset. PsA psoriatic arthritis; PsC cutaneous-only psoriasis; QC quality control