Cleaning up data

data <- data_raw %>%
  select (time, Well, BS, TF, Nar, Replicate, Abs1:Fluo2) %>%
  rename (OD = Abs1, FL_B = Fluo1, FL_T = Fluo2, Time = time)

data$Nar <- as.factor(data$Nar)
data$Replicate <- as.factor(data$Replicate)

med <- data %>%
  filter (BS == "EZ rich")

data <- data %>%
  mutate(ODc = OD - mean(med$OD), FL_Tc = FL_T - mean(med$FL_T), FL_Bc = FL_B - mean(med$FL_B)) %>%
  mutate (RFU_T = FL_T/OD, RFU_B = FL_B/OD, RFU_Tc = FL_Tc/ODc, RFU_Bc = FL_Bc/ODc)

nomed <- data %>%
  filter (BS != "EZ rich")

Basic plots

All data

Time = 30h seems a good timepoint for stationary phase analysis

Only medium

mean(med$OD)

## [1] 0.0983114

mean(med$FL_T)

## [1] 37.62719

mean(med$FL_B)

## [1] 61.01974

grid.arrange(m2, m3, m1, nrow = 2)

medium OK.

Without medium

Defining biosensor subsets

control <- nomed %>%
  filter(BS == "Control")
BS_a <- nomed%>%
  filter(BS == "AB001")
BS_b <- nomed %>%
  filter(BS == "AB002")
BS_c <- nomed%>%
  filter(BS == "AB003")
BS_d <- nomed %>%
  filter(BS == "AB004")

Check per biosensor strain

Control

Seems that Nar has effect on growth (also in following graphs) - nicer spreading of results for the bottom measurements compared to the top (RFU_B vs RFU_T): larger dynamic range so trends better visible (for example horizontal trend for exp phase)

BS_a

(Warning because scale was adapted for plot)

## Warning: Removed 1 rows containing missing values (geom_point).

BS_b