Curve fit and KD recovery

Hello. I ran an experiment in which my Ag levels were 1.25, 2.5, 5.0, 10.0, 20nM. vs. a mAB. I noticed the curve has slightly poor fitting using all points. If I take out the 20nM, I get a better fit.. The sensorgram with all concentrations had 5.55E-11 M, while the sensorgram without the 20nM came out to be 3.99E-11 M. Which one would be better?
Thanks.

Hi,
I think that the fittings are quite good. In our group we consider these two values not different (~40-60 pM). Also, you reported precision with two significant digits is too optimistic in my view.
The highest concentration of 20 nM is > 300x the K_D. High analyte concentrations compared to the K_D can suffer from deviations of the 1:1 model. I your case this is luckily very minimal. Second, your lowest analyte concentration is still above the K_D. I know that with these very low dissociation rates it is difficult to get curvature in the lowest concentrations. You can make the association time longer to get more curvature.
I would not put too much weight to the deviation at 20 nM of analyte. As you encountered, leaving out curves gives you a different answer. Even a second fit on the same curves can give you slightly different values. When you perform this experiment again you get probably also different kinetic values.
To get a better estimate of the kinetics you should perform the experiment two or three times, ideally with different batches of analyte. If you want to do it perfectly, you should also use different ligand batches and ligand densities.

Kind regards
Arnoud

thanks Arnoud for your reply