Element quantifications
From Ben Doughty, via Email:
For the effect size calculations, since we do a 6-bin sort, we don't actually compute a log2-fold change. Instead, we use the data to compute an effect size in "gene expression" space, which we normalize to the negative controls. The values are then scaled, so what an effect size of -0.2 means is that this guide decreased the expression of the target gene by 20%. An effect size of 0 would be no change in expression, and an effect size of +0.1 would mean a 10% increase in expression. The lowest we can go is -1 (which means total elimination of signal), and technically the effect size is unbounded in the positive direction, although we never see _super_ strong positive guides with CRISPRi.