Dose–Response & IC50 Lab
Potency is not efficacy, and the curve proves it
Fit sigmoidal dose–response curves, compare compounds head to head, add experimental noise, and compute a therapeutic index against a toxicity curve.
The takeaway
A compound can be ten times more potent and still be the worse drug. Potency moves the curve left; efficacy sets how high it goes — and only one of those decides whether the patient gets better.
R = Bottom + (Top − Bottom) / (1 + ([D]/IC50)^h) · signal falls with dose
Log-dose response curve
% activity remaining vs log₁₀[drug]Plotted on a log x-axis the hyperbola straightens into a sigmoid, and the IC50 lands exactly at the inflection. Sliding a curve LEFT is potency. Raising or lowering its PLATEAU is efficacy. They are independent — and confusing them is one of the most expensive mistakes in pharmacology.
Live verdict · potency vs efficacy
| Compound | IC50 (potency) | Max effect (efficacy) | Hill slope | Fold vs best |
|---|---|---|---|---|
| BNX-101 | 12.0 nM | 96% | 1.00 | 6.0× |
| BNX-207 | 400 nM | 99% | 1.90 | 200.0× |
| BNX-330 | 2.0 nM | 48% | 0.70 | — |
Potency ≠ efficacy. BNX-330 is the most potent (IC50 2.0 nM) — it works at the lowest dose — but it tops out at only 48% effect. BNX-207 needs 200.0× more drug, yet reaches 99% — it is the more efficacious molecule. If the biology demands near-complete target suppression, the potent compound is the wrong drug at any dose. Potency is a dosing convenience; efficacy is whether the drug can do the job at all.
Experimental noise & curve fitting
Turn this on to scatter simulated replicate wells around the selected compound's curve, then watch a four-parameter fit try to recover the parameters you set.
Therapeutic index · efficacy vs toxicity
Every drug has two dose–response curves: the one you want and the one you do not. Their separation on the x-axis is the safety margin. TI = TD50 / ED50.
A steep toxicity curve is dangerous: the gap between 'fine' and 'harmed' collapses.
Therapeutic index
8.0
Moderate window
Certain safety factor
0.02
TD01 / ED99 — overlap!
Usable band
799 nM → 18.7 nM
ED99 up to TD01
What IC50 really means
IC50 is not a property of the molecule. It is a property of the molecule in your assay: change the substrate concentration, the enzyme concentration, the incubation time or the cell line, and the number moves. A competitive inhibitor assayed at high [S] will look weak; the same compound at low [S] will look potent. This is why IC50 values from different papers are almost never directly comparable, and why the Cheng–Prusoff correction exists to translate an IC50 back into the assay-independent Ki.
Potency vs efficacy vs affinity
- Affinity (Kd) — how tightly the drug sticks to the target. Thermodynamics. Measured by binding, not function.
- Potency (IC50/EC50) — how much drug the system needs to hit half-effect. Depends on affinity, but also on receptor reserve, cell permeability and assay design.
- Efficacy (Top/Emax) — what the drug can achieve once it is bound. A full agonist and a partial agonist can share identical affinity and wildly different ceilings.
A drug with picomolar affinity and 20% efficacy is a beautifully bound failure.
Why this is THE core readout
Every screening cascade in the industry funnels into this curve. A hit is a molecule with a real sigmoid; a lead is one whose sigmoid keeps moving left across a chemical series; a candidate is one whose efficacy curve is far enough from its toxicity curve to survive a human being. Structure–activity relationship campaigns are, quite literally, thousands of these plots stacked on top of each other.
And a shallow Hill slope (h < 1) or a stubbornly high floor is often the earliest hint that a compound is hitting something other than the intended target.