Hello everyone. I tried measuring the concetrations of my protein fractions using eppendorf biophotometer. All the values were accompanied by the statement ''The coefficient of determination for the regression evaluation of the standard series is < 0.8''. Does that mean the out put values are wrong and I cant trust them...??
I would like to point out that the standards were prepared for a plate reader , which I used previously, as per the instructions from Thermo for their Pierce™ BCA Protein Assay Kit. Do I have to consider changing my standards when I change equipment..? Awaiting the valuable inputs and thanks a lot in advance.
Hi Phani,
Coefficient of determination or R squared represents how well the regression line, also known as model or best fit line, represents your data. In other words if R squared value is 0.8 its means that the model can explain 80% of the variations in the data surrounding the best fit line. The value comes from squaring r which is the coefficient of correlation describing the association between X and Y. That being said, if r-squared was 0.81 then r is 0.9 representing a correlation between x and y close to perfect (perfect is r=1). Depending on how accurate your measurement needs to be, usually anything above r-squared of 0.6 is considered a good fit and above 0.8 is a very good fit. Accordingly, I'd trust the regression model, but if your experiment crucially requires accurate protein concentration then I'd say redo your standards until you get a determination coefficients >0.8.
Hi Ahmed,
Thanks a lot for the explnation. I will redo the standards and I will also run a blot and normalize to see if I can trust the concentrations. Much appreciated.
Hi Phani,
can you share the data? There can be several different reasons for the R2 being low. In some cases you can also use a polynomial regression instead of a linear regression to account for the curvature observed in the dose-response curve. It is however important to understand whether the curvature is due to the assay or due to the protocol.
Cheers Markus
Hi Markus,
I have think the problem is that I prepared my samples for a plate reader procedure while its different for a spectrophotmeter based measurement. And also, I measured the concentrations of both standards and the samples as you would measure the RNA concentrations that is by placing a couple of microliters of stad/lysate onto cuvette and measuring the concetration after blanking with water, which I thnk is a mistake. However, I have attached the file. Thanks a lot.
Phani
Hi Phani,
as you say your problem is that you have no readings at all and you need to
use the correct protocol. Probably you have further diluted the samples and are thus below the limit of detection / linear range of the assay.
Cheers Markus
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression