Forum: help

Help analysing data [ Reply ]
By: Alex C. on 2019-09-17 15:01

2019-09-17 (3).png (86) downloads

Hi everyone,

I'm new to R and I'm trying to learn it as I think it's a very useful and powerful tool for data analysis. However, sometimes I find it a bit tricky, hence why I joined this community. So below is a description of the data that I need to analyse.

In a study, individual bacterial cells were examined. They contained a plasmid with a GFP gene fused to a promotor that senses antibiotic induced damage. The goal was to see if the severity of the stress response would correlate with bacterial survival when exposed to antibiotics.

Bacteria were stressed, but not killed, with low antibiotic concentrations. GFP fluorescence intensity was measured by taking the median signal intensity over each cell. Subsequently, these same cells were exposed to high concentrations of the antibiotic and after 260 minutes it was registered whether (`Death` = 1) or not (`Death` = 0) they died.

The experiment was conducted four times. Cells within one experiment were grown in the same batch and ended up on the same microscopy slide. Maybe on one day the stove was opened more often, or the bacteria in some batch always have higher expression of the operon of interest. In any case, bacteria within one experiment, and thus those on the same slide, are more correlated with each other than with cells from another experiment, or slide. Therefore, it would be good to account for this nesting of cells within slides by means of a random effect.

Each row in the data set is an observation on a single bacterial cell. So for each cell, the data set contains values for the following variables (columns):

* `Slide`: the slide ID; on which slide every cell was located.
* `GFP`: the median fluorescence intensity per cell.
* `Death`: whether (1) or not (0) the cell died.
* A whole bunch of other variables informative on each cell (covariates).

So I need to perform an analysis to test whether the stress response intensity (`GFP` signal strength) correlates with bacterial survival (`Death`), this while accounting for the random effect of the slide.

So my question is, what kind of model do I need? Is this a binomial GLM? If so, I'm not quite sure what the workflow for this is.

Any help will be greatly appreciated. Thanks!