Help! What stats should I use?

#1
Hello, Can someone advise me on what stats test should I apply here?
I work with a certain microorganism and I am treating the cells with X compound for 5 cycles (creating a phenotype that is different from its parent). After the cycles are finished I take a sample of those cells (which I call them as adapted) and plate them on solid media containing lethal concentrations of the compound X to determine survival (log CFU/ml). I also run control (or parent or non-adapted) cells along with and similar to the adapted cells. I repeat this 3-4 times on different days. I tried using an unpaired two-tailed t test to determine significance between the Non adapted and adapted cell types. What other stat methods can be applied here? The purpose is to determine significance between the NonAdapted and Adapted cells. Can ANOVA be used for data like such? My knowledge of Stats is lacking and I am still learning basics of it in school. Please help. Thanks in advance.
 
#2
It sounds alot like a minimum inhibitory conc. (MIC) type experiment? That is are you using a dilution series of compound X onto the cells? Might be better to think about it that way.

But If just comparing the CFUs then consider the following fun methods spotted in a laboratory near you:
a) Log transform (log cfu/mL) and t-tests-away. I'd say that is good to go like 90% of the time. Standard t-test caveats apply, see package insert for details. and in addition, you need the counts to not be too close to 0, preferably not close at all, preferably staying in the sweet spot of ~100CFU before dilution correction, and obviously 0 is a non-starter. Report the geometric mean and 95% CL by taking anti-log of various quantities arising from log transformed data.

b) if you have alot of 0's or you need to use small counts for some reason, and probably it is best to avoid this by manipulating the experiment to, say reduce the concentration of 'Compound X', if that is its real name, you may be needin to hit it up with some poison regression or related procedure. This is going to be good time to make friends in Stats department, but it is not beyond the ken of mere mortals.


Here is a link to some dude doing similar analysis on COVID plaques. Same-difference right? github.com/dylanmorris/sars-cov-2-stability. Don't worry though its not this complicated he just did bayesian super-complicated way, you know how people are...

Could also run non-parametric.

Well gotta go feed my cat.