Statistical analyses
Statistical analyses were performed by using the statistical language R (www.R-project.org). The quality of samples was assessed using FastQC. Each sample was aligned to the human reference genome by using STAR (open-source aligner).40 Mapped sequencing reads were assigned to genomic features using the featureCounts function. Counts were transformed to log scale by voom-transform and fit in a linear model.41 Fold-changes (FCHs) were estimated, and hypothesis testing was conducted by using contrasts under the general framework for linear models in the limma package. P-values were adjusted for multiple hypotheses by using the Benjamini-Hochberg procedure, controlling for false discovery rate (FDR). Genes with FCH≥2 and FDR<0.05 were considered DEGs.