In picousse/gRowthAnalysis:

How to use:

library(devtools)
load_all()

library(furrr)
plan(multisession, workers = availableCores()-1)

data input.

XML file

code underneath is able to parse the tecan xml output into a rather simple csv file that can then be straightforwardly processed. This function is not so fast, and thus it is best to do this once at the beginning of an analysis, write to a csv, and work from the csv latter on.

file = "data/20190618_bidir_GFPmorange_GFPgain95/2019-06-07 15-56-57_plate_1.xml"
xml.in = read.tecan.xml(file)

write.csv(xml.in, "xml.parsed.csv", row.names = FALSE)

df = readr::read_csv("xml.parsed.csv")

Metafile

The current concept starts with calculations per well, and only at the final point add the meta information (with a simple join). Summaries can easily be made using the summarize function from dplyr.

Data cleanup/restructure

We'll split up the dataframe to retrieve only the core data. While doing so, we will also condense the data.

# start calculations
df.temperature <- df %>%
  filter(variable == "Temperature")

# condens table
df.1 <- df %>%
  filter(!(variable == "Temperature")) %>%
  # time.to.hour("time") %>%
  group_by(well, variable) %>%
  nest()

Calculations

Expontential time

df.exp.time <- df.1 %>%
  filter(variable == "ABS_600_600_NA") %>%
  mutate(exp.time = map_dbl(data, ~calc.exp.time(., input_time = "time"))) %>%
  ungroup() %>%
  select(well, exp.time)

# join with the original table
df.exp <- df.1 %>%
  left_join(df.exp.time)

# create dataframe with data only from exponential time
df.2 <- df.exp %>%
  mutate(data.exp = map2(data, exp.time, ~.x %>% filter(time < .y)))

Mu / q

Calcualte q's. mu is special case of q.

df.q <- df.2 %>%
  select(well, variable, data.exp) %>%
  mutate(fit = map( data.exp, ~q.calc.individual(.)))

ratio's

df.tmp <- df.2 %>% filter(variable == "ABS_600_600_NA") %>% ungroup()
df.tmp2 <- df.2 %>% filter (variable != "ABS_600_600_NA") %>% ungroup()

df.ratio <- df.tmp2 %>%
  left_join(df.tmp %>% select(norm.data = data, well), by= c("well" = "well")) %>%
  mutate(data =  map2(data, `norm.data`, ~left_join(.x, .y, by = c("time" = "time"), suffix= c("", ".norm")))) %>%
  select(-data.exp ) %>%
  select(-norm.data) %>%
  mutate(data.exp = map2(data, exp.time, ~.x %>% filter(time < .y)))

df.ratio.fit <- df.ratio %>%
  mutate(lin.fit = map(data.exp, ~ratio.calc.individual(., x.signal = 'value.norm', y.signal='value'))) 

o <- ratio.extract.data(df.ratio.fit, "lin.fit")