eorePrep.R

# DATA PREP
# For: Shiny demo of Eore Radial Tree
# Trying to emulate: https://observablehq.com/d/e4c0d5b05b440e32 (Shweta's diagram)
# See also: https://rdrr.io/cran/networkD3/src/R/radialNetwork.R

library(shiny)
library(tidyverse)
library(networkD3)
library(data.tree)
library(readxl)

# Data preparation (comment out)
# Read in Shweta's file
eoreoutput_vsr <- read_excel("eoreoutput_vsr.xlsx")
eore.df <- eoreoutput_vsr %>% select("solute_name_initial_1",         # Level 1
                                     "total_system_pressure_initial", # Level 2
                                     "temperature_initial",           # Level 3
                                     "solute_name_initial_2",         # Level 4a
                                     "solid_name_initial_1",          # Level 4b
                                     "solute_concentration_initial_2", # Level 5a
                                     "solid_mass_initial_1",           # Level 5b
                                     "max_polymer_length_final_3",     # Level 6
                                     "paper_title",
                                     "paper_author",
                                     "paper_year"
                                     )

eore.df <- data.frame(lapply(eore.df, function(x) {
  gsub("--", NA , x)
}))

# Change `max_polymer_length_final_3` to numeric (for selection)
eore.df$max_polymer_length_final_3 <- as.numeric(eore.df$max_polymer_length_final_3)

# Append every non-NA row of Level 2
eore.df$total_system_pressure_initial[!is.na(eore.df$total_system_pressure_initial)] <- paste0(eore.df$total_system_pressure_initial[!is.na(eore.df$total_system_pressure_initial)], " bar")

# Append every non-NA row of Level 3
eore.df$temperature_initial[!is.na(eore.df$temperature_initial)] <- paste0(eore.df$temperature_initial[!is.na(eore.df$temperature_initial)], " C")

# Append every non-NA row of Level 5a, 5b
eore.df$solute_concentration_initial_2[!is.na(eore.df$solute_concentration_initial_2)] <- paste0(eore.df$solute_concentration_initial_2[!is.na(eore.df$solute_concentration_initial_2)], " mmol/l")
eore.df$solid_mass_initial_1[!is.na(eore.df$solid_mass_initial_1)] <- paste0(eore.df$solid_mass_initial_1[!is.na(eore.df$solid_mass_initial_1)], " mg")

# Find Level 4 multiples
lev4mults <- which(!is.na(eore.df$solute_name_initial_2) & !is.na(eore.df$solid_name_initial_1))

# replicate those rows (at the end)
for (i in lev4mults) {
  eore.df <- rbind(eore.df,eore.df[i,])
  eore.df$solute_name_initial_2[i] <- NA
  eore.df$solid_name_initial_1[nrow(eore.df)] <- NA
}

# Find Level 5 multiples
lev5mults <- which(!is.na(eore.df$solute_concentration_initial_2) & !is.na(eore.df$solid_mass_initial_1))

# replicate those rows (at the end)
for (i in lev5mults) {
  eore.df <- rbind(eore.df, eore.df[i,])
  eore.df$solute_concentration_initial_2[i] <- NA
  eore.df$solid_mass_initial_1[nrow(eore.df)] <- NA
}

# Combine Level 4's
eore.df$Level4 <- apply(eore.df[,4:5], 1, function(x) paste(x[!is.na(x) ], collapse = ""));
eore.df$Level5 <- apply(eore.df[,6:7], 1, function(x) paste(x[!is.na(x) ], collapse = ""));

# Reorganize
eore.df <- eore.df %>% select(solute_name_initial_1,
                              total_system_pressure_initial,
                              temperature_initial,
                              Level4,
                              Level5,
                              max_polymer_length_final_3,
                              paper_title,
                              paper_author,
                              paper_year)

colnames(eore.df) <- c("Level1",
                       "Level2",
                       "Level3",
                       "Level4",
                       "Level5",
                       "Level6",
                       "paper_title",
                       "paper_author",
                       "paper_year")

eore.df$pathString <-paste("RNA Experiments",
                           eore.df$Level1,
                           eore.df$Level2,
                           eore.df$Level3,
                           eore.df$Level4,
                           eore.df$Level5,
                           eore.df$Level6,
                           sep="|")

# Remove the first entry
eore.df <- eore.df[-1,]

# Save it!
saveRDS(eore.df, "eore.df.Rds")