Calibration curve

[Gamewizard]

Hi,

 

I need to plot a graph with mean absorbance values against concentrations. Basically there would be 4 mean absorbance values on the y-axis and concentrations would be on x-axis. And then I need to use this information to determine the concentration of my protein (RFP) in unknown sample. I wanted to do this on excel but i dont know how to do this bit, and also on paper. Does anyone know how to do this ?

 

Thanks

[ewmon]

Is the "curve" linear?

[aliu]

It sounds like you are trying to do quantitative analysis for an ELISA analysis with a standard curve with 4 points to interpolate the concentrations of your unknown samples. If so, you will need to do a curve fit for your standard points. The most popular curve fit model to use for immunoassays is the 4 parameter logistic or 4PL which is a nonlinear regression model. MasterPlex ReaderFit is software specifically designed for ELISA/dose response quantitative analysis. If you can send me your absorbance values for your standards and unknowns (along with your standard concentrations), I would be more than happy to do a sample analysis for you. If you would like to try it for yourself, I would like to invite you to download the free 14-day fully functional trial of MasterPlex ReaderFit.

 

Please feel free to ask me any questions you may have.

 

* Disclaimer: I am an employee of Hitachi Solutions.

 

Allen Liu

Supervisor, Technical Support

MiraiBio Group of Hitachi Solutions

[ewmon]

If your standards are in the linear portion of the curve (ie, log of concentration versus absorbance is linear), then you can use a linear trendline in Excel along with the curve's parameters to work backwards from absorbance to concentration.

 

You can also interpolate between two neighboring standard points ... that is, the ones that are closest to, and on either side of, the unknown's absorbance. Because you're also doing this on paper, you may not need to perform a sophisticated curve fit (such as the 4PL method) for the solution using a computer.

[Gamewizard]

Is the "curve" linear?

 

Yes it is supposed to be straight.

 

spam removed

 

Yes you are right I am doing ELISA. But I am not supposed to use a special software for this work, I can either do it on excel 2007 or by hand.

 

If your standards are in the linear portion of the curve (ie, log of concentration versus absorbance is linear), then you can use a linear trendline in Excel along with the curve's parameters to work backwards from absorbance to concentration.

 

You can also interpolate between two neighboring standard points ... that is, the ones that are closest to, and on either side of, the unknown's absorbance. Because you're also doing this on paper, you may not need to perform a sophisticated curve fit (such as the 4PL method) for the solution using a computer.

 

Okay so in excel, would that be the normal trendline ? but then how would I calculate the concentration ?

[ewmon]

Most likely, it'll be a straight plot on a log-linear graph, where the concentration is along the log axis and the absorbance is along the linear axis.

 

1. In Excel, plot the concentration along the x axis as a logarithmic scale, and the absorbance along the y axis as a linear scale.

 

2. Add a logarithmic trendline, and choose the options for displaying the R² value and the equation on the chart. The trendline's R² shows how well it fits the known data points (ie, the standards), and it should be close to 1 (ie, let's say 0.95 or better, depending on the requirements of the assay) in order to be a valid assay.

 

3. Solve the equation for x (ie, Conc), plug the Abs value(s) of the unknown into the equation, and compute its concentration.

 

4. Plot the same standards on log-linear graph paper. (If you don't have any, or don't want to buy a whole pad of it for this one experiment, then there's free graphs on the Internet. Try to find one that's as big as possible to give you the greatest accuracy. However, don't use a graph with too many log cycles because that'll make the line unnecessarily "shallow", which will degrade its accuracy. I mean, try to find a graph with only the number of log cycles you require.)

 

5. Eyeball the best fit with a straight-edge and draw it in. You'll be surprised how well the human eye can estimate the best fit. Mark the unknown's Abs value on the y axis, draw a line horizontally to the best fit line, then draw a line vertically to the x axis, and then mark and read the corresponding Conc value on the x axis.

 

6. Good luck with computing the R² value from the graph, if you need to do so. (I've been too spoiled by the microplate reader software and GMP requirements, so I've never calculated it from a paper graph.)

Edited March 12, 2011 by ewmon