Dear all,
I have a question regarding the most appropriate fit of a standard curve in an enzyme-linked lectin assay set-up to quantify the amount of mucin secretion in conjunctival explant cultures. I want to compare the following three conditions; (I) basal mucin secretion in their normal culture medium, (II) secretion after a starvation period, and (III) intracellular mucin content (obtained through a freeze-thaw cycle).
We recently discovered that the optical density of our standard curve differs when made in the three media related to the three conditions. We not only obtained different values, the resulting standard curve also had a different type of curve as best fit. For example, one standard curve benefits from a linear curve while the standard curve in a different media shows an exponential curve as best fit. The question now is if it is appropriate to just use the best fit for each standard curve within the experiment or should we use one fit (for example linear or polynomial curve grade 4) for the three standard curves?
Thank you in advance,
If the different behaviors of the standard curves is reproducible, then I think you should use the fitting method that is best for each curve, rather than the same method for all three.
Adam B Shapiro, thank you for your answer! The different behaviors of the standard curves are reproducible.
Try logistic regression. The five parameter version (5PL) fits a sigmoid curve with assymetric inflections. Bascially that means an "S" shape where the convex and concave ends have different radii. The advantage of this equation is that it can model variability in dose-response e.g. saturation of optical density, or other factors, allowing a better fit at the ends of your calibration range. There are also weightings you can use e.g. 1/Y to improve fit at low concentrations, or whatever is appropriate to your experimental design.
If you are observing that different conditions produce different degrees of linearity in your dose-response, maybe try to fit some 5-PL curves and see how the five constants that define each equation are changing for each dataset.
Kindly check https://www.ncbi.nlm.nih.gov/pmc/articles/PMC270131/
Also check https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6050442/
Karl De Ruyck, thank you for the suggestion! I never tried 5-PL curves but I will definitely explore this option to see if I obtain a better fit.
- Badges
- Science method
- Similar topics
- Medicine
- Immunology
- Immunology Techniques
- Immunoassay
- ELISA