Experimental data analysis

[matiasandina]

write functions to aggregate movement data and plot

• [x] Get all movement data to the same baseline before aggregation

Baseline can be established within animal within light stage, as the mean of the 10% lowest responses. Then subtract this to the raw movement before aggregation into temporal bins.

Likely Useful plots

Individual

• [X] line: individual movement vs time for a specific date, 24h on x axis
  • [x] x axis should be able to display ZT time and geom_rect() shades to show light/dark accordingly
• [x] heatmap: individual movement across days. Each row is a day or each row is an individual?

  Buttons to add

• [x] Display/filter days accordingly

Grouped

• [ ] group plot line with ribbon

[matiasandina]

Looks like this function can repair the lights affecting the baseline and correct everything to baseline zero. For those values below the mean of the 0.1 quantile, the movement will be negative (probably too small so it might not modify things).

repair_baseline <- function(x){
  # we take the 0.1 quantile as the lowest values
  base <- mean(x[x < quantile(x, 0.1)])
  # we return x - base
  return(x - base)
}

[matiasandina]

This app needs to get revisited and redone separate into a separate project, maybe with its own repo. Some thoughts here:

• [ ] lineplot.R has to be fixed, it's trying to handle the x axis too much. If x_axis handling has to be done, it might be better to separate it into a different function.
• [ ] light_shade.R is trying to calculate the shading, this was better done in make_lights.R on the FED quantification. The specifics might be a tad different but the logic of how to calculate the odd/even moments for the lights might be good to steal from that file.
• [x] Movement has a zero at the first point. This is a carryover of the way the optic flow algorithm works. When we repair the baseline to put things to zero, this will impact the values and become largely negative. I decided to get this "solved" with slice(-1) to remove the first row before calculating repair_baseline.

[matiasandina]

• [X] Solve the lights off lights on switch problem. Most direct way is to have a filter with a value 2e05 should be safe enough? Convert this to NA and fill.

solved by

           mutate(movement = ifelse(movement > 2e05, NA, movement),
                   movement = zoo::na.approx(movement, na.rm=F, rule=2))