Question: What should I consider when designing my 3' or 5' CRISPR pool?
Answer: When designing a CRISPR pool, some key factors should be taken into consideration:
The number of perturbations being included in the pool/study
The number of guides per perturbation
The percentage of non-targeting guides present in the pool/study
The number of perturbations (or targets) and the number of guides per perturbation are both largely driven by the specific research question to be answered.
If the experiment is an exploratory screen, we typically suggest incorporating 2-5 gRNA's per gene to identify guides with high on-target activity (i.e., knock-down or activation) and low off-target activity. This ensures that, when moving towards the final pool, the guides can be down-selected to only those with high on-target activity avoiding false positives or less efficient gRNAs.
Non-targeting guides are used as negative controls which do not recognize any sequence in the human or mouse genome. It is also vital to ensure that a proportion of cells containing non-targeting guides is included in the pool to ensure that the perturbation (i.e., knock-down or activation) can be calculated.
When designing the pool, we suggest following these general principles to enable an accurate assessment of perturbation (i.e., knock-down or activation) efficiency:
- Majority of academic publications use 2-5 sgRNAs per targeted gene. The 10x Genomics CRISPR Screening assay is not impacted by the number of sgRNAs used per gene, though recovery of sufficient cells per protospacer is important for detecting statistically significant perturbations.
- Aim for at least 100-200 cells per guide to ensure a decent ratio of cells with a perturbation to cells without a perturbation.
- Typically 2-5 control non-targeting sgRNA can be used per experiment. For the 10x Genomics CRISPR Screening assay, the exact number of control sgRNAs used isn't important.
- 500-1000 cells should contain non-targeting sgRNAs.
Products: Single Cell Gene Expression, Single Cell Immune Profiling