Massively parallel quantification of SpCas9 gRNA efficiencies by surrogate lentiviral libraries

In this study, we used a surrogate lentivirus library to capture CRISPR editing outcome in HEK cells. The dataset include quantification of indel frequencies for SpCas9 gRNAs in 12,000 surrogate sites. After filtering low quality sites, the high quality SpCas9 gRNA activities from a total of 10592 sites have been used to develop an improved deep learning-based prediction model CRISPRon (https://rth.dk/resources/crispr/.). The surrogate lentivirus library contains 12,000 (12K) synthetic DNAs, each expressing a gRNA and contain a surrogate target site. The surrogate site is designed to capture CRISPR editing efficiency and outcomes, representing the corresponding endogenous site (37 nt). The 12K surrogate lentivirus library was generated by array-based oligo synthesis, followed by inserting the oligo pool into a lentiviral vector by Golden-Gate Assembly. The 12K library was then transduced to wildtype and SpCas9 expressing HEK293T cells using a MOI of 0.3. The transduced cells were changed to selection medium (puromycin) to enrich transduced days. For the SpCas9 HEK293T cells, overexpression of SpCas9 was also induced by addition of doxycycline. Indel frequencies introduced in the surrogate sites was quantified in cells 2, 8 and 10 days after transduction by deep sequencing. As an important QC step for filtering indels introduced by oligo synthesis and PCR, the synthetic oligo pool and the 12K surrogate plasmids were also deep sequenced. Deep sequencing was performed by MGISEQ-500, pair-end 150 cycles.