[R] Fitting ELISA measurements "unknowns" to 4 parameter logistic model

[David Winsemius]

On Feb 1, 2011, at 11:08 AM, Christopher Anderson wrote:

> Hello,
>
> I am trying to fit my Elisa results (absorbance readings)  to a  
> standard
> curve. To create the standard curve model, I performed a 4-parameter
> logistic fit using the 'drc' package (ExpectedConc~Absorbance). This  
> gave me
> the following:
>> FourP
>
> A 'drc' model.
>
> Call:
> drm(formula = Response ~ Expected, data = SC, fct = LL.4())
>
> Coefficients:
> b:(Intercept)  c:(Intercept)  d:(Intercept)  e:(Intercept)
>        1.336          6.236         85.521         59.598
>
>> summary(FourP)
>
> Model fitted: Log-logistic (ED50 as parameter) (4 parms)
>
> Parameter estimates:
>
>              Estimate Std. Error  t-value p-value
> b:(Intercept)  1.33596    0.15861  8.42309  0.0011
> c:(Intercept)  6.23557    3.18629  1.95700  0.1220
> d:(Intercept) 85.52140    2.15565 39.67313  0.0000
> e:(Intercept) 59.59835    5.18781 11.48815  0.0003
>
> Residual standard error:
>
> 1.866876 (4 degrees of freedom)
>
> Now that I have the 4 parameters, how do I fit the absorbance  
> readings for
> the analytical unknowns to the standard curve model (as to estimate  
> the
> concentrations of my unknown analytical samples)?
> I can use the argument 'predict', but this predicts absorbance given
> concentrations (y given x), I need to predict concentrations give  
> absorbance
> (x given y).
>

Use approxfun() to construct a curve using the predict() values as x  
at reasonable intervals and the y values you used for predict. It  
effectively swaps the inputs to predict x from y. There is a worked  
example from two weeks ago in the Archives with the Subject line:  
"Inverse Prediction with splines"


> Thanks!
> Chris
>


David Winsemius, MD
West Hartford, CT