01Identify the competition model before using the equation

GraphPad’s one-site Ki model assumes a single class of sites, reversible binding and equilibrium. The tracer concentration and its Kd enter the relationship between the competitor’s Ki and the observed curve midpoint.GraphPad — One-site competitive binding: fitting Ki (opens in a new tab)

For the simple competitive case with negligible depletion, the relationship can be written Ki = IC50 ÷ (1 + [L]/Kd), where [L] is the tracer concentration on the appropriate free-concentration basis and Kd belongs to that tracer.GraphPad — One-site competitive binding: fitting Ki (opens in a new tab)

Do not insert the competitor’s unknown affinity into the denominator. The two ligands have different roles in the calculation, even when their reported quantities share the same concentration units.

This article concerns a defined binding model. It does not provide a general conversion for any reduction in a cell response, enzyme signal or viability measurement labelled IC50.

02Work through the correction with explicit units

In an original hypothetical example, let IC50 be 30 nM, tracer concentration be 2 nM and tracer Kd be 1 nM. The denominator is 1 + 2/1 = 3, giving Ki = 10 nM.

The concentration ratio in the denominator is unitless. If one input is entered in micromolar and the other in nanomolar without conversion, the arithmetic no longer represents the stated model.

The calculation does not create a new independent measurement. Its result depends on the fitted midpoint and the supplied tracer parameters. Their uncertainty and appropriateness remain part of the interpretation.

03Different midpoints can be consistent with one Ki

Now retain the hypothetical Ki of 10 nM and tracer Kd of 1 nM. At a tracer concentration of 1 nM, the model predicts IC50 = 20 nM. At 4 nM tracer, it predicts IC50 = 50 nM.

The midpoint changed even though the assumed competitor affinity did not. This constructed comparison explains why a raw IC50 ranking across differently configured binding experiments can be misleading.

Conversely, identical IC50 values measured with different tracer conditions need not imply identical Ki values. The missing information cannot be recovered from the midpoint alone.

04Check the reasons a simple correction may not apply

GraphPad’s radioligand-analysis guidance explains that ordinary competition analysis relies on free ligand remaining approximately equal to the added concentration. Substantial ligand depletion requires an analysis that addresses that difference.GraphPad guide to radioligand binding data — Ligand depletion (opens in a new tab)

A report should therefore state the tracer identity, affinity evidence, concentration basis and model. It should also explain whether the observed curve supports the assumed interaction rather than treating a successful fit as proof of mechanism.

If those inputs are missing, retain the reported IC50 with its assay conditions and describe Ki as unresolved. Guessing a tracer Kd to complete the spreadsheet creates false precision.

A sound comparison distinguishes measured observations, fitted curve parameters and calculated affinity estimates. Keeping that chain visible is more useful than converting every value into a single apparently uniform column.