This data has all of the normalization steps already applied. Normalized read counts. Essentially we used a method out of the R package edgeR to normalize for both read depth and composition, followed by log2 transformation, and then subtraction of PCA1.