High-throughput screening (HTS) tests hundreds of thousands to millions of compounds against a biological target in a short period. It’s a foundation of small-molecule drug discovery and increasingly used in chemical biology research.
The HTS workflow
- Identify and validate target
- Develop and validate the assay
- Adapt to high-density format (384- or 1536-well)
- Pilot screen on a subset to optimize conditions
- Run primary screen across the full library
- Confirm hits with replicate testing
- Run dose-response on confirmed hits
- Triage by chemistry and orthogonal assays
- Hit-to-lead optimization
Assay formats
Biochemical assays
Use purified protein and a defined substrate. Examples: enzyme activity assays, binding assays (FRET, TR-FRET, AlphaScreen). Clean signal but may miss compounds that depend on cellular context.
Cell-based assays
Use intact cells. Capture cellular permeability, off-target effects, and pathway-level responses. Examples: reporter gene assays, viability assays, phenotypic assays.
Phenotypic screens
Look for a desired phenotype (e.g., cell death, differentiation, marker expression) without specifying the molecular target. Slower to optimize but identify novel mechanisms.
Imaging-based screens (high-content screening)
Capture multi-parameter readouts via automated microscopy. Can quantify many features per cell — morphology, localization, intensity. Used in image-based phenotypic screens, toxicity profiling, and CRISPR arrays.
Library composition
| Library type | Size | Purpose |
|---|---|---|
| Diversity library | 50K–2M | Broad chemical space coverage |
| Focused library | 1K–50K | Target-class enriched (kinases, GPCRs) |
| Fragment library | 1K–10K | Small fragments for FBDD |
| Bioactive library | 2K–10K | Drugs and tool compounds |
| DNA-encoded libraries | 10⁶–10¹¹ | Affinity selection at huge scale |
Automation and infrastructure
HTS depends on robotics: liquid handlers, plate readers, plate movers, environmental control, and integrated software. Modern facilities can screen ~100,000 compounds per day. Quality depends on careful calibration, scheduled maintenance, and consistent reagent lots.
Assay quality metrics
- Z′-factor: Statistical measure of assay quality. Z′ ≥ 0.5 is required for HTS; ≥ 0.7 is excellent
- Signal-to-background ratio
- Plate uniformity: Edge effects from evaporation are a common issue
- Day-to-day variability: Tracked with control wells on every plate
Hit identification and triage
Initial screen identifies primary hits — typically 0.1–1% of the library. These go through:
- Cherry-pick confirmation: Re-test in triplicate
- Dose-response: Determine IC50 or EC50
- Counter-screens: Eliminate compounds active against unrelated targets (PAINS, fluorescence interference, aggregation)
- Orthogonal assays: Test in a different format to confirm mechanism
- Chemistry triage: Filter out compounds with poor drug-likeness or known PAINS substructures
Common pitfalls
- Pan-assay interference compounds (PAINS): Reactive groups, redox cyclers, fluorescent compounds give false hits in many assays
- Aggregation artifacts: Many small molecules form colloidal aggregates that nonspecifically inhibit enzymes — Triton or DTT can disperse them
- Edge effects: Evaporation in outer wells causes systematic bias
- Plate effects: Reagent stability, temperature gradients, position-dependent biases
- Hit attrition through chemistry: Many initial hits fail at confirmation or counter-screen — plan for 90%+ attrition
Beyond traditional HTS
- Virtual screening: Computational methods evaluate billions of molecules at lower cost
- DNA-encoded libraries (DELs): Test 10⁶–10¹¹ molecules in a single affinity selection
- Phenotypic screens with omics readouts: Drug-induced gene expression signatures, multi-parameter cellular profiling
- Cell-Painting and Cell Profiler: Image-based phenotypic profiling of compounds
HTS is most successful when assay quality, library curation, and triage strategy are tightly controlled. The technology is mature but the planning still distinguishes successful campaigns from expensive misses.
