Perturbing one gene tells you whether that gene matters; perturbing every gene at once tells you which genes matter for the process you care about. Genome-wide disruption has moved from random insertional mutagenesis — gene traps, transposons, chemical mutagens — to programmable CRISPR libraries, and each generation has changed what questions are tractable. The unifying logic is selection: introduce a library of perturbations, apply a pressure, and let sequencing report which perturbations survived. The analysis is a population-genetics problem disguised as a molecular biology one.
Gene traps and transposons: the pre-CRISPR toolkit
Gene trapping uses a reporter or selection cassette with a splice acceptor and a promoterless marker. When the cassette inserts into an intron, it captures the endogenous transcript and simultaneously disrupts it, creating a fusion that reports insertion and detects the mutation in a single step. The classical application was large-scale screens in mouse embryonic stem cells, and the resulting resources remain a reference for mouse gene function.
Transposon systems — piggyBac, Sleeping Beauty, and the Tol2 system used in zebrafish — insert a cassette at quasi-random sites and can be mobilised to create new insertions or excise. Their advantages for screens are high efficiency, the ability to generate libraries of insertions, and reversibility (excision can revert a phenotype, which is strong evidence that the insertion caused it). Their disadvantage is that insertion is random, so coverage is uneven and many genes are hit in only one allele; generating a homozygous disruption requires breeding or serial transposition, and insertions in essential genes are inherently hard to recover.
Random mutagenesis (ENU in mice, chemical mutagens in zebrafish) produces point mutations, including hypomorphs and temperature-sensitive alleles, which can reveal functions that a complete knockout cannot. It requires mapping: the mutation must be localised by linkage analysis, positional cloning, or sequencing, and this step is the bottleneck. Modern whole-genome sequencing has made mapping much faster, which is why organismal mutagenesis screens remain viable.
CRISPR sgRNA library design
A genome-wide CRISPR screen introduces a pooled library of single-guide RNAs, each targeting one site in the genome, together with Cas9. The design decisions determine the screen's sensitivity and specificity.
- Targeting window: guides that cut early in the coding sequence — within the first 50–70% of the open reading frame — are more likely to produce a functional knockout. Exon length and position matter: cutting in a short exon that is constitutively spliced is more disruptive than cutting in an alternatively spliced exon.
- Library size: typical libraries carry 4–10 guides per gene. More guides per gene increase statistical power but dilute sequencing depth per guide; a library of 80,000 guides at 500× coverage needs 40 million reads per sample.
- Specificity: guides are scored for predicted off-target activity, usually by counting near-matches elsewhere in the genome and weighting by their position and mismatch type. Tools such as CRISPOR, the Doench scoring rules, and the Rule Set 2 model predict on-target efficiency; exact rules have been revised over time and should be applied as reported for the specific Cas9 variant and species.
- Primary screen versus secondary validation: design the library for the primary screen with many guides per gene, then validate the top hits with individual guides and, ideally, a second orthogonal guide set or an independent knockout strategy.
For non-coding screens, the design problem changes: instead of disrupting a protein, the goal may be to delete a regulatory element (paired guides flanking an enhancer, often called a CRISPR-deletion or PEP screen) or to modulate transcription with CRISPR interference (CRISPRi) or activation (CRISPRa), which are catalytically dead Cas9 fusions that repress or activate rather than cut. These approaches avoid double-strand breaks and are less toxic, but they produce knockdown or overexpression rather than null alleles, which changes the interpretation.
Running the screen and analysing enrichment
A selection screen compares guide abundance before and after a pressure. Enrichment screens (positive selection) ask which guides become more abundant — resistance to a drug, survival of a stress; dropout screens ask which guides disappear — genes required for proliferation or viability. The readout is sequencing of the guide cassette, and the analysis is a comparison of guide counts across time points relative to a reference.
- Coverage: maintain at least 200–500 cells per guide at every step, including the bottleneck of the selection itself. Loss of coverage is the most common cause of a noisy screen.
- Replicates: independent transductions, not technical replicates, are required for statistical significance. Three independent replicates is the minimum for a credible hit list.
- Analysis tools: MAGeCK and its variant-based extension (MAGeCK-VISPR), and the more recent statistical frameworks that model guide-level and gene-level effects, compute a per-gene score and a false-discovery rate. Compare against the negative-control guides to calibrate the null distribution.
- Frequent hitters: genes that appear in many unrelated screens (often components of growth-factor signalling, chromatin, or essential machinery) are context factors rather than specific hits, and should be flagged, not reported as discoveries.
The output of a screen is a ranked list of candidate genes. A screen on its own establishes that a perturbation changes the population under a specific selective pressure; it does not establish mechanism, directness, or relevance to the physiological process. Validating the top hits individually, ideally with orthogonal perturbations, is what turns a screen into a result.
Off-target effects and what to do about them
Cas9 can tolerate mismatches outside the seed sequence, producing off-target editing at sites that resemble the guide. The three common mitigations are computational (choose guides with no close genomic matches), experimental (validate the top hits with multiple independent guides, ideally targeting different regions of the same gene), and direct (measure off-target editing by whole-genome sequencing, GUIDE-seq, CIRCLE-seq, or DISCOVER-seq).
For pooled screens, the powerful internal control is the requirement that multiple guides to the same gene show concordant effects: a true hit is one in which independent guides agree. Even this is not airtight, because a shared off-target site can produce concordant effects if the guides share a seed region, which is unlikely with well-designed libraries but not impossible. High-fidelity Cas9 variants, truncated guides, and the use of nickase pairs reduce but do not eliminate off-target activity, and their reduced on-target efficiency must be compensated by longer selection times or higher coverage. Reporting the Cas9 variant, the library version, and the guide-level statistics is the minimum standard for a screen that others can reproduce.
References
- Skarnes WC, Auerbach BA, Joyner AL. A gene trap approach in mouse embryonic stem cells: the lacZ reported is activated by splicing, reflects endogenous gene expression, and is mutagenic in mice. Genes Dev. 1992;6(6):903–918.
- Zambrowicz BP, Friedrich GA, Buxton EC, et al. Disruption and sequence identification of 2,000 genes in mouse embryonic stem cells. Nature. 1998;392(6676):608–611.
- Ivics Z, Hackett PB, Plasterk RH, Izsvák Z. Molecular reconstruction of Sleeping Beauty, a Tc1-like transposon from fish, and its transposition in human cells. Cell. 1997;91(4):501–510.
- Wang W, Lin C, Lu D, et al. Chromosomal transposition of PiggyBac in mouse embryonic stem cells. Proc Natl Acad Sci U S A. 2008;105(27):9290–9295.
- Cong L, Ran FA, Cox D, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;339(6121):819–823.
- Shalem O, Sanjana NE, Hartenian E, et al. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science. 2014;343(6166):84–87.
- Wang T, Wei JJ, Sabatini DM, Lander ES. Genetic screens in human cells using the CRISPR-Cas9 system. Science. 2014;343(6166):80–84.
- Doench JG, Fusi N, Sullender M, et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat Biotechnol. 2016;34(2):184–191.
- Li W, Xu H, Xiao T, et al. MAGeCK enables robust identification of essential genes from genome-scale CRISPR/Cas9 knockout screens. Genome Biol. 2014;15(12):554.
- Tsai SQ, Zheng Z, Nguyen NT, et al. GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat Biotechnol. 2015;33(2):187–197.