reshape_xray.r

library(dplyr)
path.xray.out <- './df_xray_long.csv'

# Columns to be retrieved from rgb data and added to xray data
col.add <- c('POT_BARCODE', 'TREATMENT', 'VARIETY', 'DFP', 'GROWTH_STAGE')

# File path to rgb dataframe to retrieve TREATMENT, VARIETY, DFP, GROWTH_STAGE
path.rgb.long <- './df_rgb_long.csv'

# File path pattern containing xray results
pattern <- paste0('/depot/smarterag/data/XRAY/',
                  'Experiment_Bayer2/',
                  'Experiment_*/',
                  'kimimaro_out_200/',
                  '*.txt')

# Get file paths containing individual phenotyping results
files <- Sys.glob(file.path(pattern))

# Open individual files and merge them into a dataframe
xray <- data.frame()
for (f in files){
    tmp <- read.csv(f)
    xray <- rbind(xray, tmp)
}




# Change column names
names(xray)[names(xray)=='experiment_ID'] <- 'EXP ID'
names(xray)[names(xray)=='plant_ID']      <- 'POT_BARCODE'
names(xray)[names(xray)=='RESOLUTION']    <- 'View'

# Remove unnecessary columns
xray <- subset(xray, select=-c(pot_height))




# Load rgb dataframe (long format table)
rgb.long <- read.csv(path.rgb.long)

# Get attributes per each POT_BARCODE (TREATMENT, VARIEY, DFP, GRWOTH_STAGE)
attributes <- rgb.long %>%
               group_by(POT_BARCODE) %>%
               summarise(TREATMENT=names(which.max(table(TREATMENT))),
                         VARIETY=names(which.max(table(VARIETY))),
                         DFP=names(which.max(table(DFP))),
                         GROWTH_STAGE=names(which.max(table(GROWTH_STAGE)))) %>%

               ungroup()

# Add more attributes for consistency with rgb and hsi data
#attributes$SCAN_TIME <- -1
#attributes$SCAN_DATE <- -1
attributes$frame_nr  <- -1

# Add the attributes to xray data
xray <- left_join(xray, attributes, by = 'POT_BARCODE')

# Reshape xray dataframe
id.vars.xray <- c("EXP ID", "POT_BARCODE", "VARIETY", "TREATMENT", "DFP", 
                  "GROWTH_STAGE", "View", "frame_nr")

xray.long <- reshape2::melt(xray, id.vars = id.vars.xray, 
                            variable.name = "variable")

# Export dataframe
write.csv(xray.long, path.xray.out, row.names=F)


# EOF