Dear all,
one of my colleagues asked me a curios question. She would like to know what is the experimental variability of MTT/MTS assay on human cells because she is creating an in silico models to predict LC50 and she would like to compare the variability of her predicted values with the experimental values. I answered that in my research I accept CV minor than 10% when I work with 4 replicates so i think that the LC50 derive from my diose-response assay could have a CV minor than 10%. But I have no reference about it.. The professor who teach me how to work with cells told me that acceprability threshold for replicates is 10%.
Can someone help me and my colleague??
Thank you in advance
Considering the relatively weak predictive power of in silico models (especially for something as complex as inhibition of cell growth), 10% variability in the experimental measurements is definitely low enough to support the correlation attempt. Even a variation of 2-fold would be adequate for the purpose.
Dear Diego,
Within a given experiment based on sextuplicates, we estimate that the SEM should not exceed 10%, and we always target 5%
From one experiment to another with the same cell line, you can face higher SEM between mean values because the GI50s largely depend on the cell concentration you seed, the cell concentration before you detach them for a new seeding, etc..., etc..
If you want to have a pretty good idea about GI50's variability, you have to look on the NCI public website providing you with tousands of GI50s relating to thousands of compounds. For many compounds, you have several experiments that have been performed over time.
The NCI is using the SRB assay with an incubation time of cells - compound of 48h.
We are using the MTT assay with an incubation time of cell - compound of 72h.
The NCI already helped us several times with new compounds: the differences in mean GI50s between the NCI and our own values were < 10%.
Best regards
Robert