In npjc/growr: High level functions to help summarise growth curve fits.

knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(readyg)

Overview

In my trying to understand how ACCESS generates its numbers I had to understand what they call the od generation point... what a mess

files <- list.files('~/data/yeast-grower-db/db/01', full.names = T)

d <- tibble(file = files,
       content = map(files, read_yg)) %>% 
    unnest(content)
run_params <- d %>% 
    filter(!map_lgl(run_params, is.null)) %>% 
    unnest(run_params)

What are all the different types and values of OD generation point from a sample of r length(files) yeast grower files?

run_params %>% 
    filter(str_detect(key, 'od_gen_pt')) %>% 
    count(key, value)

There is only one for 96 wells: od_gen_pt_96 = 0.46
For 48 well plates it is most often set to 0.76 but sometimes 0.46 .. this seems like a mistake carrying over the value from 96 well leading to serious overestimation of the average amount of time per generation...

From the ACCESS text:

$OD_5$ generations, the calibrate OD equivalent to five generations, also called the OD generation point, is indicated by (*). The time interval for G_by_interval is calculated back four doubling (4G) from $OD_5$ generations (*).

Avg_G is the average generation time from start to the OD generation point and includes both lag and exponential phases in its calculation
G_by_interval calculates the doubling time of the exponential growth phase by excluding the first doubling from the calculation and estimates lag phase.

fix this bb

Get all the files that are 96 well runs:

files_of_96 <- run_params %>% filter(key == 'plate_size', value == '96')
d96 <- d %>% 
    semi_join(files_of_96)
d96_data <- unnest(d96, data)
d96_run_params <- unnest(d96, run_params)
d96_data %>% 
    group_by(file, well) %>% 
    summarise(min_value = min(value),
              max_value = max(value)) %>% 
    filter(!is.na(min_value), min_value < 0.5) %>% 
    ggplot() +
    geom_density(aes(min_value))

If we look at the run paramters, there are only two kinds of plate specifications provided (greison or nunc) and the defined 'od tecan equivalent to 5 generations of growth IF your initial suspension is od595)

d96_run_params %>% spread(key, value) %>% select(plate_size, plate_type, gens_to_od_pt_96,od_gen_pt_96) %>% distinct()

Shake mode is always 'Orbital'
Shake intensity is almost always 'high'
Shakre duration is almost always '800,0' this isn't documented anywhere but I assume that is shaking duration in seconds since 800 seconds is 13 minutes and 20 seconds and the usual measurement interval is 15 minutes (900 seconds).

filtered_set96 <- d96_run_params %>% 
    spread(key, value) %>% 
    filter(shake_mode == 'Orbital',
           shake_intensity %in% c('high', 'High'),
           shake_duration == '800,0') %>% 
    distinct(file)

The $OD_{tecan}$ for $OD_{biophotometer} = 0.0625$ seems to be ~ 0.0625

filtered_d96_data <- d96 %>% 
    semi_join(filtered_set96) %>% 
    unnest(data)

filtered_d96_data %>% 
    group_by(file, well) %>% 
    summarise(min_value = min(value),
              max_value = max(value)) %>% 
    filter(!is.na(min_value), min_value < 0.3) %>%
    ggplot() +
    geom_density(aes(min_value)) + 
    geom_vline(xintercept = 0.0625, lty = 'dashed', size = 0.5) + 
    scale_x_continuous(labels = c(0, 0.0625, 0.1, 0.2, 0.3), breaks = c(0, 0.0625, 0.1, 0.2, 0.3))