I have been performing dose response experiments on cancer and control patients for cell counts, apoptosis (flow cytometry DAPI-AnnexinV staining), intracellular protein staining on flow cytometry and cell cycle. As I have 5 concentrations of drugA (+control=6) ± drugB (as it is fixed concentration the combination might be treated as a different group).
From what I have seen so far, most biologists would compare the results using ANOVA or pairwise comparisons. However, I am not completely happy with that approach as I feel that I lose information. I have been using the drc package for calculating the IC50 of the drugs I have used (LL.4) and I was thinking in using the function EDcomp for comparing the different IC50s (I understand the package performs a t-test). Doing this I understand that I use more information in each test than using ANOVA-pairwise t.tests but I am still not sure if it is the ways for accounting for most of the information (comparing the slopes?).
Also, I have used this for the cell counts but I was planning to use it also for the other experiments I have performed as far as I have enough points (for progenitor colonies I have 4 points for one set of drugs and 3 for another set, so the model does not seem to fit well). Would it be right to use this modelling for calculating other data than cell counts? At the end of the day I am just calculating another kind of dose-response, right?
Do you have any opinion about this?