Script to load ethoscope data using rethomics and generate plots and tables that quantify behavior and make it easy to select the flies that sleep rebound the best

2024-10-10_metadata.csv

constants.R

id_labeller <- function(id, with_date = FALSE) {
  if (with_date) {
    paste(
      substr(id, 1, 10),
      substr(id, 22, 23),
      substr(id, 28, 29)
    )
  } else {
    paste(
      substr(id, 22, 23),
      substr(id, 28, 29)
    )
  }
}

SUMMARY_TIME_WINDOW_MINS <- 30
VELOCITY_CORRECTION_COEF <- 0.0048
TIME_WINDOW_LENGTH <- 10
MIN_TIME_IMMOBILE <- 300
ETHOSCOPE_DATABASE <- "/ethoscope_data/results"
ETHOSCOPE_CACHE <- "/ethoscope_data/cache"
ZT_END_OF_EXPERIMENT <- 60 # ZT when later data is not needed in the plots

paramters.yaml

filename: "2024-10-10_metadata.csv"
loading_dates:
    - "2024-10-10"

sleep_rebound_selection.R

library(data.table)
library(scopr)
library(ggetho)
library(ggprism)
library(yaml)
theme_set(ggprism::theme_prism(base_size = 20) + theme(
  text = element_text(size = 10)
))
source("constants.R")
PARAMETERS_FILE <- "parameters.yaml"
stopifnot(file.exists(PARAMETERS_FILE))
parameters <- yaml::read_yaml(PARAMETERS_FILE)
METADATA_FILENAME <- parameters$filename
LOADING_DATES <- parameters$loading_dates

# Load metadata and link to database
# Metadata must have the following fields:
# region_id: Integer identifying the position in the arena where the fly was loaded
# date: Character describing the day when the fly was loaded in an ethoscope,
#   in GMT time and format YYYY-MM-DD
# machine_name: Name of the ethoscope where the fly was loaded, typically ETHOSCOPE_XXX
# reference_hour: Numeric describing when does the fly experience ZT0, in GMT
# motor: One of YES, NO, or FAIL depending on whether the fly was intended to be sleep deprived or not.
#   If it was intended, whether it was succesful or not
# sex: Character describing what is the sex of the fly. Recommended: MAL, FEM and VFEM (for virgins)
# Other useful fields you want to have in your metadata but not required by this code
# genotype
# food
# schedule


metadata <- data.table::fread(METADATA_FILENAME)
metadata[, date := as.character(date)]
metadata[, region_id := as.integer(region_id)]
metadata <- metadata[
  date %in% LOADING_DATES &
    motor == "YES",
]
metadata <- scopr::link_ethoscope_metadata(metadata, result_dir = ETHOSCOPE_DATABASE)
metadata[, date := substr(as.character(datetime), 1, 10)]
metadata[, fly_no := paste0(substr(machine_name, 11, 13), "_", region_id)]

# Load ethoscope data using linked metadata
dt <- scopr::load_ethoscope(
  metadata,
  reference_hour = NA,
  verbose = TRUE,
  cache = ETHOSCOPE_CACHE,
  FUN = sleepr::sleep_annotation,
  time_window_length = TIME_WINDOW_LENGTH,
  velocity_correction_coef = VELOCITY_CORRECTION_COEF,
  min_time_immobile = MIN_TIME_IMMOBILE
)

# Quantify behavior in bins of SUMMARY_TIME_WINDOW_MINS minutes
dt_bin <- behavr::bin_apply_all(
  dt,
  y = "asleep",
  x_bin_length = behavr::mins(SUMMARY_TIME_WINDOW_MINS),
  FUN = mean
)

# Project rebound data on baseline for nice visualization
df0 <- behavr::rejoin(dt_bin)
df0_baseline <- data.table::copy(df0)
df0_baseline[, trace := "Baseline"]

df0_rebound <- data.table::copy(df0)
df0_rebound[, t := t - behavr::days(1)]
df0_rebound[, trace := "Rebound"]
df <- rbind(df0_baseline, df0_rebound)
df[, trace := factor(trace, levels = c("Baseline", "Rebound"))]


# Plot per fly
options(repr.plot.height = 10, repr.plot.width = 30)
gg <- ggplot(
  data = df[t < behavr::hours(ZT_END_OF_EXPERIMENT)][
    (trace == "Baseline") | (t >= behavr::days(1) & t < behavr::days(1.5)),
  ],
  mapping = aes(x = t, y = asleep * 30, linetype = trace)
) +
  stat_pop_etho(aes(color = sex)) +
  scale_y_continuous(name = paste0("Sleep per ", SUMMARY_TIME_WINDOW_MINS, " min bin (min)"), limits = c(0, 30), breaks = seq(0, 30, 5)) +
  stat_ld_annotations(height = 1, color = NA, alpha = 0.2) +
  guides(linetype = "none") +
  scale_x_hours(name = "ZT", breaks = behavr::days(seq(0, 2.5, 0.5))) +
  facet_wrap("id", labeller = labeller(id = id_labeller))
suppressWarnings({
  print(gg)
  ggplot2::ggsave(
    plot = gg,
    filename = "plot_per_fly.svg",
    width = 30, height = 10
  )
})

# # Plot per sex
options(repr.plot.width = 20, repr.plot.height = 6)
gg_sd <- ggplot(
  data = df[t < behavr::hours(ZT_END_OF_EXPERIMENT), ][
    (trace == "Baseline") | (t >= behavr::days(1) & t < behavr::days(1.5)),
  ],
  mapping = aes(x = t, y = asleep * 30, linetype = trace)
) +
  stat_pop_etho(color = "black") +
  scale_y_continuous(
    name = paste0("Sleep per ", SUMMARY_TIME_WINDOW_MINS, " min bin (min)"),
    limits = c(0, 30), breaks = seq(0, 30, 5)
  ) +
  stat_ld_annotations(height = 1, color = NA, alpha = 0.2) +
  guides(linetype = "none") +
  scale_x_hours(name = "ZT", breaks = behavr::days(seq(0, 2.5, 0.5)))
gg <- gg_sd + facet_grid(sex ~ .)

suppressWarnings({
  print(gg)
  ggplot2::ggsave(
    plot = gg,
    filename = "plot_per_sex.svg",
    width = 20, height = 6
  )
})

Put all three files in a directory in a computer with an ethocope database. Modify the parameters.yaml file to load the flies you want and execute sleep_rebound_selection.R

The provided 2024-10-10_metadata.csv file is the example metadata that generates the example figures here