IC50 calculator
Fit a four-parameter logistic curve to concentration-response data. Computed as you type.
Fit a four-parameter logistic curve to concentration-response data. Computed as you type.
The IC50 is the concentration of an inhibitor that reduces a response to half of its maximal inhibition. It is the standard way to compare the potency of drugs, antibodies and enzyme inhibitors: a lower IC50 means a more potent compound. Because it depends on substrate concentration, cell density and incubation time, an IC50 is only comparable between experiments run under the same conditions.
Here x is the concentration and y the response. Top and Bottom are the upper and lower plateaus, the IC50 is the midpoint on the log-concentration axis, and theHill slope sets how steep the transition is. A slope of 1 is a simple one-site curve. This is the same model used by GraphPad Prism's "log(inhibitor) vs. response – variable slope".
Responses are averaged per concentration, then all four parameters are estimated by non-linear least squares (Levenberg–Marquardt) on the log-concentration scale. Starting values come from the data extremes and a log-logit line, and the fit is refined until the sum of squared residuals stops improving. R² reports how much of the response variation the curve explains. The per-point table shows each observed mean against its fitted value so outliers are easy to spot.
EC50 is the same midpoint for a stimulatory curve. If your responses rise with concentration, the fit still works and the Hill slope comes out negative. Ki, the inhibition constant, is an intrinsic binding affinity that does not depend on assay conditions; for competitive inhibitors it can be derived from an IC50 with theKi to IC50 converter (Cheng–Prusoff equation).
One concentration per line, followed by one or more response values. Separate columns with spaces, tabs or commas, so a paste straight from a spreadsheet works. Extra response columns are treated as replicates and averaged. Concentrations must be above zero; a "0" control cannot sit on a log axis, so leave it out or enter it as a very small value. Use at least four concentrations, with points on both plateaus.
By fitting a four-parameter logistic (4PL) curve to response versus log concentration. The fit finds the bottom and top plateaus, the Hill slope, and the concentration at the midpoint between the plateaus, which is the IC50. This calculator does the full non-linear least-squares fit in the browser.
At least four concentrations, and ideally six to ten spanning the whole transition, with points on both plateaus. Replicates at each concentration improve the estimate; enter them as extra columns and they are averaged.
It describes how steep the transition is. A slope near 1 is the classic single-site curve. Slopes much larger than 1 suggest cooperativity or a threshold effect; slopes well below 1 suggest heterogeneous binding or a shallow, drawn-out response.
IC50 is the concentration that gives half of the maximal inhibition; EC50 is the concentration that gives half of the maximal effect for an activator. Both come from the same 4PL fit, only the direction of the curve differs. IC50 depends on assay conditions, whereas Ki is a true binding constant.
Because the data do not define both plateaus, so the fit extrapolates. Add lower or higher concentrations until the responses flatten at both ends, then refit.