Keep all the fits from growthcurver on a plate

growthcurver_example.R

# load in packages
library(growthcurver) 
library(tidyverse)

# load in example data
d <- growthdata                          # load some sample, simulated data
gc_fit <- SummarizeGrowthByPlate(d)  # do the analysis
plot(gc_fit)                             # plot your data and the best fit
gc_fit$model %>% broom::augment()

gc_fit$vals

# want to grab the fit of all the wells in the plate so we can look at their output together

# setup output dataframe. This is the key step, want every individual curve in there
output <- tibble(well = colnames(d)[colnames(d) != 'time']) %>%
  mutate(model = list(NA),
         vals = list(NA),
         data = list(NA))

# run a for loop to do each fit
for(i in 1:nrow(output)){
  well <- output$well[i]
  gc_fit <- SummarizeGrowth(d$time, d[, well])
  
  output$model[[i]] <- gc_fit$model
  output$vals[[i]] <- gc_fit$vals %>% unlist() %>% bind_rows() %>% mutate(across(k:auc_e, as.numeric))
  output$data[[i]] <- gc_fit$data
}

# get predictions for each well
d_preds <- mutate(output, preds = map(model, broom::augment)) %>%
  unnest(preds) %>%
  select(-vals, - data, -model)

# plot these
ggplot(d_preds) +
  geom_point(aes(t, n)) +
  geom_line(aes(t, .fitted), col = 'red') +
  facet_wrap(~well)

# get summary of fits for each well
d_check <- mutate(output, model_check = map(model, broom::glance)) %>%
  unnest(model_check) %>%
  select(-vals, - data, -model)

# get the values of all the desired traits
d_traits <- unnest(output, vals) %>%
  select(- data, -model)

head(d_traits)