Why most hits are lies
A primary screen produces a hit list, and most of it is noise. A false positive is a compound that looks active but is not acting on the target the way you hoped. The classic culprits are worth knowing by name. Aggregators form tiny colloidal particles that sequester protein and inhibit almost everything non-specifically. Reactive compounds chemically modify the protein indiscriminately. Some molecules absorb or emit light at the assay's wavelength and fool the detector without touching the protein at all.
These mechanisms are insidious because they produce beautiful, potent-looking dose responses. A compound can give a clean IC50 and still be doing nothing real to your target. This is why a number alone never settles anything; you have to ask how the compound is producing that number, not just how big it is.
PAINS and structural alerts
PAINS — pan-assay interference compounds — are substructures that show up as hits again and again across unrelated screens. Their promiscuity comes from the bad behaviours above: redox cycling, covalent reactivity, metal chelation, fluorescence. Curated PAINS filters flag these substructures automatically, and a flagged compound deserves serious extra scrutiny before you trust it.
A disciplined triage
Triage is the craft of turning a long, dirty hit list into a few series worth real chemistry. The goal is to fail bad hits fast and cheap, so that scarce effort goes only to molecules that can survive. The order below is roughly cheapest-and-most-discriminating first.
- Reconfirm potency from freshly weighed, re-purified solid — not the original plate sample. Many "hits" evaporate here.
- Run the detergent test and a counter-screen to expose aggregators and assay-interference artefacts.
- Apply PAINS and structural-alert filters; investigate, do not auto-delete, anything flagged.
- Confirm binding by an orthogonal, mechanism-aware method — biophysics or a target-engagement readout.
- Cluster survivors by scaffold and look for early SAR and good ligand efficiency; carry forward only series, not lone molecules.
The deepest test of a real hit is interpretable SAR: when small, sensible changes to the molecule move potency in sensible directions, you are almost certainly engaging a real binding site. A lone potent compound with no analogues and a flat or chaotic SAR around it is exactly the kind of thing that wastes a year. Honest triage is what converts a raw hit list into the trustworthy leads the next track will optimize.