Protocol Online logo
Top : New Forum Archives (2009-): : ELISA and Immunoassay

ELISA test - set up and data analysis - (Dec/09/2011 )

I'm a new member of forum and a new user of ELISA tes. I have some questions for you.

O.D. values of a reader is the same of another reader or not? Are they similar? Can I compare O.D. valuese of a reader with the values obtained with other readers?

I will use a commercial kit for dosage a protein.

I will follow the instructions on the data shee of the kit, but how do I know if the values obtained ​​of the calibrators are correct (O.D. values?) or not?
Can I understand this thing with data analysis? I will use the 4 parameter logistic model. There is a parameter usefull for undestand if the calibration curve is ok?

Thank you very much for your help.

Paolo<h3 class="r"></h3>


no, you cannot directly compare od values from one machine to another. they may be in the ballpark but don't expect them to be the same.

you can plot the results of the calibrators to determine if they worked okay.

and, you compare your samples to the calibrators. the result of this will be comparable across machines and plates.


Thank you very much for your answer. So I will dosage my samples using a commercial kit. After I will analyse data with 4 parameters logistic model. I will verify if my calibrators have worked in the right way. Could you suggest me which is the statistic valued that I will observe? R values? And it's good values range?
Thank you



I would not worry about the 'commercial kit' calibrators working...if they did not produce a dose response curve no one would be buying the kit. The question is does the kit provide accurate values; values equivalent to another method preferably a reference or accepted method or values that you would expect based upon spike and recovery study.

The only way you can 'compare' your samples to the 'calibrators' is if you know the values of the samples. You are also assuming the matrix of the calibrators and the samples is the same and that you have no matrix effect.

You are looking for r to be as close to 1 as possible with an intercept of 0 and a slope of 1. You should also determine %CV of the test by running several replicates of your first postive calibrator and your last. %CV should be < 5%.

OD values may not be the same across instruments but you can normalize the data to see how they compare.


Hello! I have used my kit for a test. I'd like to ask you some suggestions.

I obtained the follow calibration values:


I have also obtained values ​​for unknown samples.

The first step will be to build the calibration curve obtained with the values ​​and consider the statistical values ​​of output (using 4 parameters logistic model). If the statistical values ​​are good I will have to proceed with the calculation of the concentrations of unknown samples.
Is this method right?

Help me, please

Thank you very much.



Firstly your 'calibration values' are actually optical densities at specific concentrations of analyte. You construct your curve and interpolate the values of your unknown samples from your 4 PL curve. I don't know what you mean by 'statistical values are good' before you calculate your unknowns. Just go forward.

Your statistical values are comparision of your values to any reference values (another method) or by the concentrations you spiked into the matrix.
Basically observed v. expected. Reference method v. new method. etc... other things you can do are:
spike and recovery into a specific matrix
linearity (serial dilution of high sample v. high calibrator) to check for matrix effects. curves here should be parallel.
finally specificity among other tests...

The curve results are nicely spaced and cover a 2 log change. Now how do your results compare to what you exepect?



I obtained that Abs value
I constructed my curve: RMSE 0.0149 and Rsquare 0.9999 using calibrators values
I interpolated my unknows values with curve and calculated protein concentrations using as reference.

In my next test I will use calibrators, samples in double.
I wanted to ask what are the values ​​of % CV that I should consider.

Thank you


%CV should be </= 5%. Very high concentrations and low concentrations will have a higher %CV. Run 5 reps at your midpoint.

In general %CV plot (y axis) v. Concentration (x axis) should look like a "U".