Fitting sigmoid doseresponse curves with Prism

curvefit.com was created by GraphPad Software, Inc. Send comments or questions to the author of these pages, Dr. Harvey Motulsky, president of GraphPad Software.

In April 2003, GraphPad released Prism 4 and published Fitting Models to Biological Data using Linear and Nonlinear Regression. This book includes all the information that comprises curvefit.com, and much more. You can read this book as a pdf file.

Fitting sigmoid dose-response curves with Prism

To fit a sigmoidal curve with Prism, the X values must be logarithm of concentration. If you entered concentrations, Prism can transform those values. Click Analyze, choose Transformations from the list of Data Manipulations, and choose to transform X to log(X). Also check the option box to create a new graph.

Since the log(0) is undefined, you cannot enter a concentration of zero as a logarithm. If you enter a concentration of zero and then transform to logarithms, the result will be blank. Instead of entering a dose of zero, enter a low concentration -- one or two log units below your lowest concentration.

Beware of entering the zero value as a value much lower than two log units below your lowest concentration. When fitting a sigmoid curve, Prism sets the initial estimated value of the IC50 to be halfway between the lowest and highest X values. If you enter a really low value for the zero concentration, the initial value will be much lower than the true IC50, and Prism may not be able to complete the curve fit (unless you override the initial value).

Before fitting a dose-response curve, make these decisions:

Decision	Discussion
Choose standard slope or variable slope?	Prism offers two equations. One uses a standard slope factor of 1.0; the other fits the slope factor. If you have plenty of data points, choose the equation with variable slope. If data are scanty, use the standard slope.
Set Top to a constant value?	Ideally, the top part of the curve is defined by at least several data points. In this case, Prism will be able to fit the top plateau of the curve. If the top plateau is not well defined by data, then you'll need to make the top plateau be a constant based on controls.
Set Bottom to a constant value?	Ideally, the bottom part of the curve is defined by at least several data points. In this case, Prism will be able to fit the bottom plateau of the curve. If the bottom plateau is not well defined by data, then you'll need to make the bottom plateau be a constant based on controls. If you have subtracted a background value, then the bottom plateau of the curve must be 0. Prism doesn't know this unless you tell it. Make Bottom a constant equal to zero in this case.
Absolute or relative weighting?	See "Weighting method" in Method options with Prism
Fit each replicate or averages?	See ""replicates" in Method options with Prism

To fit the curve: From the data table or a graph, click Analyze. Choose nonlinear regression from the list of regressions. Choose the list of classic equations and choose either "Sigmoid dose-response" or "sigmoid dose-response (variable slope)". Then click the Method and/or Constants option buttons to configure the fit according to your choices above.

Prism fits the data to the equation below. If you choose the equation with a standard slope, HillSlope is fixed to 1.0.

Y=Bottom + (Top-Bottom)/(1+10^((LogEC50-X)*HillSlope))

Why Prism fits the logEC50 rather than the EC50