Optimizing in vitro Transcribed CRISPR-Cas9 single-guide RNA Libraries for Improved Uniformity and Affordability

We describe a scalable and cost-effective sgRNA synthesis workflow that reduces costs by over 70% through the use of large pools of microarray-derived oligos encoding unique sgRNA spacers. These sub-pool oligos are assembled into full-length dsDNA templates via Golden Gate Assembly before in vitro transcription with T7 RNA polymerase. RNA-seq analysis reveals severe biases in spacer representation, with some spacers being highly overrepresented while others are completely absent. Consistent with previous studies, we identify guanine-rich sequences within the first four nucleotides of the spacer (immediately downstream of the T7 promoter) as the primary driver of this bias. To address this issue, we introduced a guanine tetramer upstream of all spacers, which reduced bias by an average of 19% in sgRNA libraries containing 389 spacers. However, this modification also increased the presence of high-molecular-weight RNA species after transcription. We also tested two alternative bias-reduction strategies: compartmentalizing spacers within emulsions and optimizing DNA input and reaction volumes. Both methods independently reduced bias in 2,626-plex sgRNA libraries, though to a lesser extent than the guanine tetramer approach. These advancements enhance both the affordability and uniformity of sgRNA libraries, with broad implications for improving CRISPR-Cas9 screens and optimizing guide RNA design for other CRISPR and nuclease systems.This is supplemental data file contains processed counts for all sequenced spacers. Each RNA-seq sample (and two DNA samples) has a corresponding csv file. Each file has three columns:seq - the sequence of the spacercount - the raw counts of the spacerclass - if this spacer is one of the ones designed (Target) or not (Mutated Spacer)An index.csv file describes the IVT parameters for each sample.