Efficient prioritization of CRISPR screen hits by accounting for targeting efficiency of guide RNA

CRISPR-Cas9 screening has been successfully used to identify novel therapeutic targets. However, the screening results are often noisy with many low-confidence hits. Therefore, correcting the effects of noise and bias in CRISPR screening data can improve the efficiency of prioritizing hits for further validation. Here, we identified the variations in target cleavage efficiency, even in optimized sgRNA libraries, that pose a strong bias in phenotype, and developed an analysis method that adjusts the phenotype score by varying the targeting efficiency of each sgRNA. Our method significantly removed bias and was better able to identify essential genes. Finally, we applied this method to identify novel genes whose ablation could synergize with vemurafenib in the A375 melanoma cell line. Our method identified nicotinamide N-methyltransferase, lactate dehydrogenase B, and polypyrimidine tract-binding protein 1 as synergistic targets whose ablation sensitized A375 cells to vemurafenib. Conventional fold change analysis was unable to identify those synergistic targets. Collectively, we expect that our new analytical method will more accurately identify genes that confer the phenotype of interest.