ELISA, What is the Best Fit for the Standard Curve?

2 posts / 0 new
Last post
sereen kh
sereen kh's picture
ELISA, What is the Best Fit for the Standard Curve?

Help!  :(

I'm working on ELISA, using ELISA kit ready coated, ELISA reader with software.

I make dilution series to create standard curve (1000 pg/ml, 500, 250, 125, 62, 31, 15) but sometimes I apply in the reaction only 5 std.

the cytokine i'm working with has very low level in human serum, almost around the lowest std. (15 pg/ml)

I got std. curve with R^2 > 0.99 which is very good but still not enough, because I need almost perfect std. curve!
where R^2 = 0.999 or even 1.0, otherwise I'll get undetectable samples conc.! so I need to delete/ignore some std. point to reach 0.999 value,  see the attachment.

I'm trying to delete only one point as much as possible using the software, but some times deletion two points (out of 5) is necessary especially the higher std. values.

My question is, is it ok if I delete 1000, 500 std. points while the sample conc. range between  0 - 250 pg/ml ? so the must important thing to get perfect linear curve within this range?

I confused & begin to wonder what are the rules to accept results!
what is the minimum acceptable R^2 value?
 Is it acceptable to delete one or two std. points?  
How can I choose the best parameters for fair results?!
When should I repeat the duplicated samples? what is the acceptable range of  the mean?

Any help will be highly appreciated.

Thank you :)

 

Sami Tuomivaara
Sami Tuomivaara's picture
sereen kh,

sereen kh,

Your standard concentrations need to cover only the expected range of or your sample(s). For example, prepare 0, 5, 10, 25 and 50 pg/ml standards if you don't expect to encounter higher values in your samples. Unnecessarily large range can introduce non-linearity or other artifacts.

Also importantly, you need to evaluate, whether your assay is reproducible enough at this range. If not, you need to tweak the assay conditions or try something else altogether (what that could be, I don't know)...

When it comes to R^2, there are no hard limits for acceptability, it is whether you accept it or don't. You can read this tutorial or google more and find lot of discussions about it, including, what R^2 really means... Then, you as a scientist make an informed conclusion based on the raw data and the R^2.

Cheers,