Hi all,
I have some calcium imaging data collected from neurons in the mouse cerebral cortex. I wish to compare the effect of two genotypes. Within each genotype, I have several neurons that were recorded for fluorescence changes over 5 mins each.
The data is therefore arranged as follows: Genotypes (main columns), individual neurons (sub-columns under each column), and time (rows).
Can someone share the best statistical method(s) to apply to compare the distributions (blue and gray lines)? I do not wish to select specific frames or perform multiple comparisons for each time frame.
Hello, Navneet.
If I understood correctly, you want some type of statistical testing to verify whether the two Fluorescence curves differ, is that right?
In this case, I think it is important to note that the two curves are not two distributions; but, rather, they are a time-varying pattern. I would say you would, in fact, have fluorescence values distributions in the data for each specific frame. In this case, I would think of performing statistical tests on each (or at a given step between frames) frame to determine the times for which you do observe differences. This should also entail careful considerations on time-dependencies, since you would be dealing with time-series.
I realize this is what you said you did not want to do. However, I am not sure whether you could apply traditional statistical testing to the time-varying patterns you plotted directly. They are not distributions per se; rather, they indicate a specific distribution for each frame.
Does that make sense?
Hi Carlos,
Thanks for your help :) Yes that's correct for your first line.
While I see what you mean by time-varying pattern and perhaps my wording was incorrect, I do not wish to perform multiple-testing for each frame.
That just tells me whether the two curves are different at individual frames but not whether the two curves are innately different.
I was wondering if a 2 way ANOVA with testing for column and row effects would be meaningful. What do you think?
I would suggest considering comparing summary values from your fluorescence transients, rather than the entire transient. In my lab, we would get an average transient in each cell, and make measurements off it, then group by experimental manipulation and compare. One obvious comparison would be the peak amplitude. To compare timecourse, you can use the width at half height (half-width), or tau by fitting to an exponential decay function. It is also common to measure the time it takes to rise or decay from 10% to 90% of the peak value.
You might get insights into the best thing to measure by scaling all the transients to the same height, and see if your blue and grey groups differ in other ways.
ANOVA sounds useful, but it only tells you two groups are different. To understand what the specific differences are, you would have to do posthoc comparisons, which leads to point by point comparisons.
Just my thoughts.
Matthew
- Badges
- Science topic
- Similar topics
- Chemistry
- Pharmacology
- Pharmaceuticals
- Drugs