Hello,
For a while, I have been conflicted on what could be the 'right' way to analyse my data. I am doing calcium imaging in individual cells using fluorescence microscopy. A stimulus is added, after which the fluorescence signal gets brighter. Ultimately, I quantify the area under the curve (AUC) of this response.
-I want to compare the AUC of two genotype: WT vs. KO.
-Cells are seeded in a container (well). In one such a well, I measure 120 cells simultaneously.
-I have measured 6 wells for each genotype.
To test whether the mean AUC between genotypes differs, I have considered pooling the AUC's of all 120 cells within a well as one mean, since one could argue one well is one biological replicate, and the individual cells are merely technical replicates. Subsequently I could do a T-test with only n=6.
However, I cannot shake this gut feeling that this is wrong, since we lose a tremendous amount of information this way. The whole single-cell aspect is lost. To retain the single-cell aspect of these measurements, I have considered fitting a mixed model "AUC ~ Genotype + (1|well)" with well measured as a random factor and perform an Anova using this model.
Is one of these two ways the "correct" way to go about this. Or am I missing a whole different aspect?