Preference of the length and sequence context of spacers in CRISPR/dCas9 inhibition (CRISPRi) and activation (CRISPRa) screens. (A) Distribution of phenotype scores (Gilbert et al. 2014) for sgRNAs targeting the top 500 essential genes and the control sgRNAs in CRISPRi experiments. The dashed line represents the threshold chosen to determine efficient and inefficient sgRNAs. (B) A bar chart showing the effect of spacer length on sgRNA efficiency. (C) Logos showing the sequence preference of spacers. The height of the nucleotides represents the log odds ratio of nucleotide frequency between efficient and inefficient sgRNAs. The nucleotide at the 5′ end of the spacers is fixed to be guanines in the library design and is excluded from the logos. (D) Bar charts comparing the performance of CRISPRi model and CRISPR/Cas9 KO model in predicting sgRNA efficiency in CRISPRi negative selection, CRISPRi positive selection upon CTx-DTA treatment, and CRISPRa negative selections in Gilbert data and Konermann data. The length of spacers is 20 nt. Cross-validation was used to assess the performance of the CRISPRi model in the CRISPRi negative selection experiment. The error bars represent the standard deviations in 10 iterations of threefold cross validation. The P-value was computed using a paired t-test.
