Table 3 Major technological adaptations to sc-CRISPR and their advantages and disadvantages
From: CRISPR screening in hematology research: from bulk to single-cell level
| Â | Adaptation | Advantages | Disadvantages |
|---|---|---|---|
Direct-capture Perturb-seq [74] | Direct gRNA capture via capture sequence or 5′ sequencing Targeted sequencing | Direct sequencing of the gRNA eliminates risk for barcode uncoupling Compatible with 3′ and 5′ sequencing Targeted sequencing reduces cost and increases scalability | Capture sequence may impact gRNA efficiency Requires specific resources compatible with direct gRNA capture Targeted sequencing is inherently biased |
Direct-seq [73] | 8A8G sequence for gRNA capture | Artificial poly-A allows poly-T-based gRNA capture Compatible with multiple different single-cell platforms Compatible with 3′ and 5′ sequencing | Requires sufficient sequencing saturation to detect gRNAs which are part of the mRNA library |
DoNick-seq [54] | Cas9 nickase in combination with pairs of gRNAs | gRNA pairs enhance knockout efficiency Reduced off-target effects | More constraints for gRNA design Risk for accidental in-frame edits Not compatible with CRISPRi or CRISPRa |
CaRPool-seq [104] | Cas13 | Cas13 targets RNA instead of DNA Processing of CRISPR array into individual gRNAs for easy gRNA multiplexing Reduced off-target effects Cas13 protein is of smaller size than Cas9 | Not compatible with CRISPRa CRISPR arrays require complex cloning strategy |
Sc-Tiling [106] | CRISPR tiling | Intragenic screening Enables identification of new protein domains Multiple gRNAs close together in the same domain create a sense of redundancy and increase power | Depending on the sequence, some domains may be more difficult to target |
Deaminase screening [108] | Base editing | Introduction of point mutations | Bias toward certain mutations |
POKI-seq [143] | Knock-in using HDR templates | Can be applied in vivo Non-viral delivery so no integration in the host genome | Knock-in may suffer from low efficiency |
(bee)STING-seq [107] | Targeting GWAS loci | Screening of non-coding regions | Screening GWAS loci tends to require large libraries with potentially little relevant hits |
In vivo Perturb-seq [113] | In vivo screening | In vivo Preserves the natural microenvironment Screening circumvents the need for establishing in vivo knockout models for each target | May suffer from poor engraftment Requires optimized tissue dissociation Requires large numbers of animals |
Perturb-map [123] | Spatial resolution | Preserves the spatial architecture of the tissue Allows analysis of tumor microenvironment | Does not reach actual single-cell resolution Number of perturbations is limited by the number of possible ProCode combinations |
Compressed Perturb-seq [103] | Computational sample demultiplexing | Allows demultiplexing in case of multiple cells per droplet or multiple gRNAs per cell Reduced cost Requires lower cell numbers Allows analysis of interaction effects as well as individual effects | Interaction effects may complicate data analysis Computational demultiplexing might generate artifacts |
TAP-seq [84] | Targeted sequencing | Requires lower sequencing depth Enables larger scale screens at a lower cost Possibility to detect lowly expressed genes | Biased Risk for poor amplification efficiency for certain amplicons |
Genome-wide Perturb-seq [136] | Genome-scale | Generates extremely rich dataset | High cost in terms of reagents and sequencing Huge data analysis effort |
PerturbSci-Kinetics [140] | RNA kinetics | 4-thiouridine labeling distinguishes nascent RNA based on T to C conversions Allows analysis of RNA dynamics (synthesis, degradation etc.) The use of combinatorial indexing does not require specialized library preparation resources and allows scaling | Treatment with 4sU may be associated with toxicity and alter physiological cell state |
Perturb-CITE-seq [132] | Proteomics | Proteomic profiling | Limited to cell surface proteins Limited number of proteins can be detected |
ECCITE-seq [131] | Proteomics | Multimodal profiling: RNA, TCR, gRNA, hashing and surface protein Hashing allows sample pooling and superloading | Limited to cell surface proteins Limited number of proteins can be detected |
Perturb-ATAC [101] | Epigenomics | Profiling chromatin accessibility | Low throughput No gene expression data |
CRISPR-sciATAC [124] | Epigenomics | Profiling chromatin accessibility The use of combinatorial indexing does not require specialized library preparation resources and allows scaling | No gene expression data |
SPEAR-ATAC [125] | Epigenomics | Profiling chromatin accessibility Improved gRNA assignment due to targeted amplification of gRNA sequences High thoughput Reduced cost | No gene expression data |