To assess differential gene expression across PCPs, we took advantage of the replicate batches within each type and performed a meta-analysis across replicates based on non-parametric statistics. Briefly, for each PCP we performed one-tailed Mann-Whitney tests between individual batches within a cell type against all cells outside of that cell type. To ensure that significance would arise from replication rather than extreme p-values, prior to meta-analysis with Fisher’s method, p-values at FDR<=0.05 for an individual test were set to the maximum p-value meeting that criterion. Finally, meta-analytic p-values were FDR corrected. Differentially expressed gene sets were defined by FDR adjusted p-value <0.05 having log2 fold change >2.
