curve-fitting, IC50 calculation

2 posts / 0 new
Last post
Marzil's picture
curve-fitting, IC50 calculation

I have evaluated Cd toxicity in different soils by root-growth tests and tried to calculate IC50 using BioDataFit. Some of my response curves are obviously not complete, they do not reach the "bottom". It is in soils where Cd is effectively immobilised and its toxicity is decreased. However, in these cases, the response, as far as I have determined it, is linear, doesn't look like the "proper" beginning of a dose-response curve. Is it possible that the response curve changes from the typical shape to a linear shape when the effect of the inhibitor is  modified by the interaction with soil components, and IC50 cannot be calculated?

Sami Tuomivaara
Sami Tuomivaara's picture


Equations used in dose-response analysis (of which growth analysis is a subset of), are derived with the assumption that the system under study can be approximated as an _equilibrium_ system. In vivo systems are inherently extremely complex and it is my impression that growth curves rarely fit well to simple model with small number of freely floating parameters (for example four: maximum growth, minimum growth, IC50 and slope). Also, as your said, some of the Cd might be slowly released from the soil or inhibititor (slower than the uptake by the plant) so your system is not in equilibrium and you get growth curve that doesn't fit well to your model. Your suggestion that the curve might be best fitted to different models in different dose regimes of the curve is absolutely correct.

When the data doesn't fit very well, you have to come up with another model to explain the behaviour. It is difficult to give unequivocal answer to your problem but first of all you have to make sure that you use correct model in BioDataFit that is appropriate for your system. It just might be that this kind of model is not part of the BioDataFit package, or that it doesn't exist yet at all... you have to write it up ;)