genome-wide CRISPR screen

Knocking out one gene tells you about one gene. But often the real question is open-ended: out of all 20,000 human genes, which ones are needed for cancer cells to survive a drug, or for a virus to infect, or for any process you care about? A genome-wide CRISPR screen answers that by doing the experiment on every gene at once and letting the result point to the important ones.

Here is the trick. You build a huge library of guide RNAs — collectively targeting every gene in the genome, several guides per gene. You deliver this library into a large pool of cells at low dose, so each cell typically receives just one guide and therefore has just one gene knocked out; the population becomes millions of cells, each missing a different single gene. Then you apply a selection or pressure — add the drug, the virus, whatever you are studying — and let it sort the cells. Afterwards you sequence the surviving (or vanished) cells to read which guide RNAs they carry. Guides that became enriched mark genes whose loss helped survival; guides that dropped out mark genes that were essential. The guide RNA doubles as a built-in barcode that tells you which gene each cell lost.

This is one of the most powerful methods in modern functional genomics, turning a hunt that once took years into a single pooled experiment, and it has uncovered drug targets, essential genes, and the machinery behind countless cellular processes. Variants use CRISPR interference or activation instead of knockout, to dial genes down or up across the genome. The honest caveats: screens find associations that must then be validated gene by gene; guides vary in how well they work; off-target effects and incomplete knockouts add noise; and a screen tells you which genes matter for the condition you tested, not the whole story of what each gene does.