I am setting up a direct ELISA and have a standard curve which I fit to a 5-PL curve. When I try to calculate the concentrations of the points in the standard curve using the obtained 5-PL curve I get very bad relative errors (>15%). Can anyone help me to understand why this is and how I can get a better standard curve.
posted numbers and concentrations will help people take a look at the data.
Hi Anette,
How far are you back-calculated concentrations of standard from the expected values? You said >15% but in the case of ligand-binding assays such as EIAs or ELISAs, and according to the EMEA (European Medicines Agency) Gudelines (2011) in enclosed files, the acceptable errors are superior: " The target back-calculated concentrations of the calibration standards should be within 20% of the nominal value (25% at LLOQ and ULOQ) for at least 75% of calibration standards. Calibration standard may be excluded from the curve due to a technical error with an assignable cause (e.g. pipetting error)." So perhaps your results are not as "bad" as you think! :)
Anyway, if they're still not acceptable taking into account these guidelines, I think you should look closer at several aspects:
- pipeting errors? Do you use manual pipete or automatic pipete?
- dilution curve error: how is prepared your standard curve? by serial dilution? if yes, are you sure there is no dilution error, I mean in the calculations or during the manipulations? Do you well mix each of your standards before pipeting in to prepare the following ones?
- What about the other reagents? Have you got some other control samples allowing you to check the reliability of your reagents?
Actually, it may be fastidious but you should review each step of your manipulation since a lot of reasons can explain the non-correctness of your standards...
Good luck,
Cécile
- Badges
- Science method
- Similar topics
- Medicine
- Immunology
- Immunology Techniques
- Immunoassay
- ELISA