Drug Target Validation: A Standardized Workflow from Hit to Model

Checklists accumulate items; workflows impose order. The difference matters because the gates in target validation are not equally expensive, and the cheap ones are only cheap if they come first.

The sequence below runs from genetic evidence (weeks, low cost) to translational models (months, high cost). Each gate has a defined question and a defined stopping rule. A programme that reaches the end has four independent lines of evidence; one that stops early has saved the cost of the rest.

The gates are ordered by cost, not by importance: gate 1 is often available from public data before any experiment, and gate 4 is where most budgets are consumed.

The strongest target validation is a human genetic association whose direction matches the intended pharmacology.

Questions to settle at this gate:

• • Is the variant associated with the disease, and in which direction? A target that lowers gene function must be protective for a loss-of-function strategy.
• • Is the gene constrained? Loss-of-function variants in the population are informative about tolerated inhibition.
• • Is the association reproducible across ancestries and cohorts?

Programmes that skip this gate can spend years on a target whose human genetics contradict the intended mechanism.

The question is not "did the knockout work" but "did the phenotype appear, and is it attributable to the target".

Design elements:

• • Modality matched to the pharmacology. An inhibitor reduces activity; a knockout removes protein. If the drug is an inhibitor, a catalytically dead knock-in or a degron is closer to the pharmacological question than a null allele.
• • Rescue. Re-expression of a resistant allele restores the phenotype; this is what separates target-specific from editing-associated effects.
• • Dose–response. Titratable systems (inducible CRISPRi) show whether the effect scales with target level — the closest in vitro analogue of a dose–response curve.
• • Cell context. A phenotype in an immortalized line may not appear in the disease-relevant cell type, which is gate 4's problem but is worth a preliminary check.

Genetic evidence can be confounded by adaptation; chemical evidence can be confounded by off-target activity. Together they are stronger than either alone.

The second row is the most decisive and the most often skipped: engineering a resistance mutation and showing the compound loses activity in that line converts a correlation into a mechanism.

This is where the validation decisions of the previous pillar apply: each field requires its own model and readout.

A programme that reaches gate 4 with only immortalized-line data typically needs another six to twelve months. Building the relevant model earlier is often cheaper than discovering at the end that the phenotype does not transfer.

EDITGENE builds the cellular evidence for gates 2 to 4: knockout, inducible knockdown, resistance-mutation lines, isogenic panels, and disease-relevant models with the field-appropriate readouts.

Tell us the target and the stage you are at. Contact EDITGENE to request a project consultation or quote.

Validation requirements differ by modality and therapeutic area; the gates described here are a general framework, not a regulatory standard.

Cellular evidence does not predict clinical efficacy. Statements should distinguish target validation from therapeutic validation, and should report the model and readout used.

• • Nelson MR, et al. The support of human genetic evidence for approved drug indications. Nature Genetics. 2015;47:856–860.
• • Plenge RM, et al. Validating therapeutic targets through human genetics. Nature Reviews Drug Discovery. 2013;12:581–594.
• • Fellmann C, et al. Cornerstones of CRISPR-Cas in drug discovery and therapy. Nature Reviews Drug Discovery. 2017;16:89–100.
• • Bunnage ME, et al. Target validation using chemical probes. Nature Chemical Biology. 2013;9:195–199.