Fig. 1

Empirical design of the HD CRISPR library. Schematic illustration of the HD CRISPR library design process. sgRNA sequences that were previously used in negative selection screens contained in the GenomeCRISPR database were annotated with information on rationally selected sequence and phenotype features. All negative selection screens were reanalyzed with BAGEL. High-quality experiments were selected based on how well reference core and nonessential gene sets could be separated in those screens. sgRNAs with high on-target activity were then determined as sequences that both target an essential gene (as determined by BAGEL) and that rank among the 20% most strongly depleted sequences in the screen. Next, sgRNAs that showed unexpected phenotypes compared to other sgRNAs targeting the same gene were flagged as outliers with potential off-target effects. sgRNA sequences were then selected from the resulting pool of sequences to design a genome-wide library consisting of two mutually exclusive sub-libraries A and B, prioritizing sgRNAs with high on-target and low off-target activity