Designing a Synthetic Lethality Screen
The design decision that comes first
| Design | What it can find | What it cannot |
|---|---|---|
| Single arm, one background | Dependencies in that background | Interactions — there is no second state |
| Two arms, isogenic pair | Interactions with the edited gene | Interactions with other genotypes |
| Panel with genotype stratification | Genotype-associated sensitivity | Causation; confounded by lineage |
| Two-arm panel | Both, at higher cost | Still needs validation |
The most common error is running a single-arm screen in a knockout background and calling the hits "synthetic lethal". A hit in that design is a gene required in the knockout; whether it is also required in the wild type was not measured.
Two-arm versus interaction scoring
A genetic interaction is defined by comparing the double perturbation with the expectation from the two singles.
| Quantity | Meaning | Requirement |
|---|---|---|
| Fitness(gene X, WT background) | Single perturbation effect | Arm 1 |
| Fitness(gene X, mutant background) | Combined effect | Arm 2 |
| Expected (multiplicative or additive) | Null model | Explicitly stated |
| Interaction score | Observed − expected | Both arms, matched conditions |
Important distinction. Multiplicative and additive null models give different answers, and the choice should be stated. For fitness readouts, the multiplicative model is conventional; for a score that saturates at zero viability, it is not appropriate and a different transformation is needed.
Library
| Library | Genes | Guides per gene | Use |
|---|---|---|---|
| Genome-wide | ~19,000 | 4–6 | Hypothesis-free |
| Druggable / metabolic subset | 2,000–3,000 | 4–6 | Cheaper, higher coverage per gene |
| Focused pathway | 100–500 | 6–10 | Mechanism, epistasis |
| Dual-guide (pairwise) | Combinatorial | — | Direct interaction mapping, very large |
For most metabolic questions the subset library is the better choice: the pathway is known, the question is which node interacts, and the coverage per gene is higher at the same cost.
Coverage arithmetic
| Parameter | Value | Consequence |
|---|---|---|
| Genes | 2,000 | — |
| Guides per gene | 5 | 10,000 guides |
| Cells per guide at selection | 300 | 3 × 10⁶ cells |
| Transduction efficiency | 30% | ~1 × 10⁷ cells at transduction |
| Doublings over the window | 8 | Coverage falls unevenly across guides |
| Reads per guide | 200 | 2 × 10⁶ reads per sample |
The third and fifth rows are where screens fail silently. Cells per guide at selection is not cells per guide at transduction, and a long window lets fast-growing clones dominate the baseline.
Readout and window
| Readout | Detects | Window |
|---|---|---|
| Viability selection | Lethality, strong sickness | 14–21 days |
| Fluorescence sorting | Marker-linked phenotype | 5–10 days |
| Expression readout (RNA) | Pathway output | 1–3 days |
| Imaging | Morphology, localisation | Fixed endpoint |
A viability screen needs enough time for the interaction to manifest and short enough for coverage to survive. For metabolic interactions, which are often graded, the window determines whether the effect is detectable at all: too short and it is absent, too long and it is masked by adaptation.
Caveat on adaptation. Cells with a partial fitness defect can acquire compensatory mutations over three weeks. A hit that disappears on re-testing in a fresh background may have been an adaptation, not an interaction.
Scoring and statistics
- • Normalise per sample for library representation before comparison.
- • Use the negative-control distribution to set the null, not the bulk of guides.
- • Model the over-dispersion. Guide-level counts are not Poisson; a negative-binomial or empirical-Bayes approach is standard.
- • Report effect size with the confidence interval. A rank is not a result.
- • Correct for the number of hypotheses, and state which correction.
Readout note. The output of a screen is a ranked list with uncertainty, not a set of hits. The cut-off is a decision about the follow-up budget, and it should be described as such.
Validation, in order
| Step | Test | Rules out |
|---|---|---|
| 1 | Individual knockout, two guides | Guide-specific off-target |
| 2 | Second, non-overlapping guide | Position effects |
| 3 | Rescue with a resistant cDNA | On-target activity |
| 4 | Compound, if one exists | Phenotype specific to CRISPR |
| 5 | Re-test in a fresh isogenic pair | Clone artefacts, adaptation |
| 6 | In vivo, if the claim is translational | Culture-specific interaction |
Steps 1–3 eliminate most false positives and are cheap. Step 5 is the one that catches the failure mode specific to long screens.
Failures that recur
- • Single-arm screen labelled synthetic lethal.
- • Null model unstated. Multiplicative and additive give different hits.
- • Coverage computed at selection only. Transduction needs several fold more cells.
- • Baseline library not sequenced. Drift is unattributable.
- • Effect size omitted. Rank alone cannot be reproduced.
- • Adaptation mistaken for interaction. Re-test in a fresh background.
- • Guide-level off-target not excluded. Two guides is the minimum.
Where this connects
- • Metabolic synthetic lethality article supplies the definitions and the evidence standard this design serves.
- • Tumour dependency article supplies the target-validation gates that follow a screen.
- • Amino acid pillar supplies the methionine and one-carbon logic behind the canonical MTAP–MAT2A interaction.
Scientific interpretation and experimental design
- • Two arms or it is not an interaction.
- • State the null model.
- • Compute coverage at transduction.
- • Sequence the baseline library.
- • Report effect size and interval.
- • Validate with two guides and a rescue.
- • Re-test in a fresh background to exclude adaptation.
Selected references for further reading
- • Hart T, et al. Measuring error rates in genomic perturbation screens: gold standards for human functional genomics. Molecular Systems Biology. 2014;10:733.
- • Horlbeck MA, et al. Compact and highly active next-generation libraries for CRISPR-mediated gene repression and activation. eLife. 2016;5:e19760.
- • Li W, et al. Quality control, modeling, and visualization of CRISPR screens with MAGeCK-VISPR. Genome Biology. 2015;16:281.
- • Kryukov GV, et al. MTAP deletion confers enhanced dependency on the PRMT5 arginine methyltransferase in cancer cells. Science. 2016;351:1214–1218.
Related Services
Related Products
| Product name | Cat.No. | Species | Gene ID | |
|---|---|---|---|---|
| PBRM1 Knockout A-549 Cell Line | EDJ0001-K01 | Human | 55193 | Details Get a Quote |
| YTHDC1 Knockout A-549 Cell Line | EDC07652 | Human | 91746 | Details Get a Quote |
| TP53 Knockout HCT 116 Cell Line | EDC07854 | Human | 7157 | Details Get a Quote |
| Sirt2 Knockout MH-S Cell Line | EDJ-KQ78173 | Mouse | 64383 | Details Get a Quote |
| TFE3 Knockout KGN Cell Line | EDJ-KQ48 | Human | 7030 | Details Get a Quote |
| Stub1 Knockout MB49 Cell Line | EDJ-KQ53 | Mouse | 56424 | Details Get a Quote |
| ARID1A Knockout SNK-6 Cell Line | EDJ-KQ64 | Human | 8289 | Details Get a Quote |
| Tet1 Knockout Kupffer Cell Line | EDJ-KQ88 | Mouse | 52463 | Details Get a Quote |
| NR1H3 Knockout HEK293T Cell Line | EDJ-KQ109 | Human | 10062 | Details Get a Quote |
| NR1H2 Knockout HEK293T Cell Line | EDJ-KQ110 | Human | 7376 | Details Get a Quote |
| RPS6KB2 Knockout HEK293 Cell Line | EDJ-KQ125 | Human | 6199 | Details Get a Quote |
| SMAD6 Knockout HEK293 Cell Line | EDJ-KQ126 | Human | 4091 | Details Get a Quote |
| JUN Knockout HEK293 Cell Line | EDJ-KQ176 | Human | 3725 | Details Get a Quote |
| JUN Knockout HEK293T Cell Line | EDJ-KQ184 | Human | 3725 | Details Get a Quote |
| ALKBH5 Knockout HEK293 Cell Line | EDJ-KQ186 | Human | 54890 | Details Get a Quote |
- 1
- 2
- ...
- Next Page »