Figure 1.

The iDEG Algorithm. 1) Normalize unequal library sizes if necessary. 2) Partition transcriptome into percentile-based windows using ranked baseline expression. 3) For each window: estimate mean expression, variance, and dispersion parameters. 4) Apply the Variance Stabilizing Transformation (VST) for each gene expression count. 5) Calculate the standard normal summary statistic “Z_g” for each gene expression count “g”. 6) Determine the identified DEG set “$\widehat{\mathcal{G}}$ ” based on a pre-determined α-cutoff.