#!/usr/bin/env Rscript
### SimText App ###
#
#The tool enables the exploration of data generated by text_to_wordmatrix or pmids_to_pubtator_matrix in a locally run ShinyApp.
#Features are word clouds for each initial search query, dimension reduction and hierarchical clustering of the binary matrix,
#and a table with words and their frequency among the search queries.
#
#Input:
#
#   1)	Input 1:
#   Tab-delimited table with
# - column with search queries starting with "ID_", e.g. "ID_gene" if initial search queries were genes
# - column(s) with grouping factor(s) to compare pre-existing categories of the initial search queries with the grouping based on text.
#  The column names should start with "GROUPING_". If the column name is "GROUPING_disorder",
#  "disorder" will be shown as a grouping variable in the app.
#  2)	Input 2:
#   Output of text_to_wordmatrix or pmids_to_pubtator_matrix, or binary matrix.
#
# optional arguments:
# -h, --help                    show help message
# -i INPUT, --input INPUT       input file name. add path if file is not in working directory
# -m MATRIX, --matrix MATRIX    matrix file name. add path if file is not in working directory
# -p PORT, --port PORT          specify port, otherwise randomly selected
#
#Output:
#Shiny app with word clouds, dimensionality reduction plot, dendrogram of
#hierarchical clustering and table with words and their frequency among the entities.
#
#Packages
if (!require("shiny") & "--install_packages" %in% commandArgs()) {
  install.packages("shiny", repo = "http://cran.rstudio.com/")
}

if (!require("plotly")  & "--install_packages" %in% commandArgs()) {
  install.packages("plotly", repo = "http://cran.rstudio.com/")
}

if (!require("DT")  & "--install_packages" %in% commandArgs()) {
  install.packages("DT", repo = "http://cran.rstudio.com/")
}

if (!require("shinycssloaders")  & "--install_packages" %in% commandArgs()) {
  install.packages("shinycssloaders", repo = "http://cran.rstudio.com/")
}

if (!require("shinythemes") & "--install_packages" %in% commandArgs()) {
  install.packages("shinythemes", repo = "http://cran.rstudio.com/")
}

if (!require("tableHTML") & "--install_packages" %in% commandArgs()) {
  install.packages("tableHTML", repo = "http://cran.rstudio.com/")
}

if (!require("argparse") & "--install_packages" %in% commandArgs()) {
  install.packages("argparse", repo = "http://cran.rstudio.com/")
}

if (!require("PubMedWordcloud") & "--install_packages" %in% commandArgs()) {
  install.packages("PubMedWordcloud", repo = "http://cran.rstudio.com/")
}

if (!require("ggplot2") & "--install_packages" %in% commandArgs()) {
  install.packages("ggplot2", repo = "http://cran.rstudio.com/")
}

if (!require("stringr") & "--install_packages" %in% commandArgs()) {
  install.packages("stringr", repo = "http://cran.rstudio.com/")
}

if (!require("tidyr") & "--install_packages" %in% commandArgs()) {
  install.packages("tidyr", repo = "http://cran.rstudio.com/")
}

if (!require("magrittr") & "--install_packages" %in% commandArgs()) {
  install.packages("magrittr", repo = "http://cran.rstudio.com/")
}

if (!require("plyr") & "--install_packages" %in% commandArgs()) {
  install.packages("plyr", repo = "http://cran.rstudio.com/")
}

if (!require("ggpubr") & "--install_packages" %in% commandArgs()) {
  install.packages("ggpubr", repo = "http://cran.rstudio.com/")
}

if (!require("rafalib") & "--install_packages" %in% commandArgs()) {
  install.packages("rafalib", repo = "http://cran.rstudio.com/")
}

if (!require("RColorBrewer") & "--install_packages" %in% commandArgs()) {
  install.packages("RColorBrewer", repo = "http://cran.rstudio.com/")
}

if (!require("dendextend") & "--install_packages" %in% commandArgs()) {
  install.packages("dendextend", repo = "http://cran.rstudio.com/")
}

if (!require("Rtsne") & "--install_packages" %in% commandArgs()) {
  install.packages("Rtsne", repo = "http://cran.rstudio.com/")
}

if (!require("umap") & "--install_packages" %in% commandArgs()) {
  install.packages("umap", repo = "http://cran.rstudio.com/")
}

if (!require("mclust") & "--install_packages" %in% commandArgs()) {
  install.packages("mclust", repo = "http://cran.rstudio.com/")
}

suppressPackageStartupMessages(library("shiny"))
suppressPackageStartupMessages(library("plotly"))
suppressPackageStartupMessages(library("DT"))
suppressPackageStartupMessages(library("shinycssloaders"))
suppressPackageStartupMessages(library("shinythemes"))
suppressPackageStartupMessages(library("tableHTML"))
suppressPackageStartupMessages(library("argparse"))
suppressPackageStartupMessages(library("PubMedWordcloud"))
suppressPackageStartupMessages(library("ggplot2"))
suppressPackageStartupMessages(library("stringr"))
suppressPackageStartupMessages(library("tidyr"))
suppressPackageStartupMessages(library("magrittr"))
suppressPackageStartupMessages(library("plyr"))
suppressPackageStartupMessages(library("ggpubr"))
suppressPackageStartupMessages(library("rafalib"))
suppressPackageStartupMessages(library("RColorBrewer"))
suppressPackageStartupMessages(library("dendextend"))
suppressPackageStartupMessages(library("Rtsne"))
suppressPackageStartupMessages(library("umap"))
suppressPackageStartupMessages(library("mclust"))

#command arguments
parser <- ArgumentParser()
parser$add_argument("-i", "--input",
                    help = "input file name. add path if file is not in working directory")
parser$add_argument("-m", "--matrix", default = NULL,
                    help = "matrix file name. add path if file is not in working directory")
parser$add_argument("--host", default = NULL,
                    help = "Specify host")
parser$add_argument("-p", "--port", type = "integer", default = NULL,
                    help = "Specify port, otherwise randomly select")
parser$add_argument("--install_packages", action = "store_true", default = FALSE,
                    help = "If you want to auto install missing required packages.")
args <- parser$parse_args()

# Set host
if (!is.null(args$host))  {
  options(shiny.host = args$host)
}

# Set port
if (!is.null(args$port)) {
  options(shiny.port = args$port)
}

#load data
data <- read.delim(args$input, stringsAsFactors = FALSE)
index_grouping <- grep("GROUPING_", names(data))
names(data)[index_grouping] <- sub(".*_", "", names(data)[index_grouping])
colindex_id <- grep("^ID_", names(data))

matrix <- read.delim(args$matrix,
                     check.names = FALSE,
                     header = TRUE,
                     sep = "\t")
matrix <- (as.matrix(matrix) > 0) * 1 #transform matrix to binary matrix

##### UI ######
ui <- shinyUI(fluidPage(
  navbarPage(theme = shinytheme("flatly"),
             id = "inTabset",
             selected = "panel1",
             title = "SimText",
             tabPanel("Home",
                      value = "panel1",
                      tabPanel("Results",
                               value = "panel1",
                               fluidRow(width = 12,
                                        offset = 0,
                                        column(width = 4,  style = "padding-right: 0px",
                                               wellPanel(h5(strong("ID of interest")),
                                                         style = "background-color:white;
                            border-bottom: 2px solid #EEEEEE;
                            border-top-color: white;
                            border-right-color: white;
                            border-left-color: white;
                            box-shadow: 0px 0px 0px white;
                            padding:3px;
                            width: 100%"),
                                               selectInput("id_interest", "Select ID:", paste0(data[[colindex_id]], " (",
                                                                                               seq(1, length(data[[colindex_id]])),
                                                                                               ")"))),
                                        column(width = 3,  style = "padding-right: 0px",
                                               wellPanel(h5(strong("Color settings")),
                                                         style = "background-color:white;
                            border-bottom: 2px solid #EEEEEE;
                            border-top-color: white;
                            border-right-color: white;
                            border-left-color: white;
                            box-shadow: 0px 0px 0px white;
                            padding:3px;
                            width: 100%"),
                                               radioButtons("colour", "Color by:",
                                                            c("Grouping variable", "Individual word")),
                                               selectInput("colour_select", "Select:",
                                                           choices = c(names(data)[index_grouping])))
                               ),
                               fluidRow(width = 12, offset = 0,
                                        column(width = 4, #style = "height:650px;",
                                               wellPanel(textOutput("id_text"),
                                                         style = "background-color: #333333;
                            color: white;
                            border-top-color: #333333;
                            border-left-color: #333333;
                            border-right-color: #333333;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-bottom: 0px;
                            padding:5px"),
                                               wellPanel(
                                                 fluidRow(
                                                   column(width = 4,
                                                          numericInput("fontsize", "Font size:",
                                                                       value = 7, min = 1, max = 50)),
                                                   column(width = 4,
                                                          numericInput("nword", "Word number:",
                                                                       value = 50, min = 1, max = 100)),
                                                   column(width = 12,
                                                          withSpinner(plotOutput("wordcloud_plot",
                                                                                 height = "325px"))),
                                                   column(width = 12,
                                                          downloadLink("download_wordcloud", "Download"))),
                                                 style = "background-color: #ffffff;
                            border-bottom-color: #333333;
                            border-left-color: #333333;
                            height: 470px;
                            border-right-color: #333333;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-top: 0px"),
                                               wellPanel(textOutput("table"),
                                                         style = "background-color: #333333;
                            color: white;
                            border-top-color: #333333;
                            border-left-color: #333333;
                            border-right-color: #333333;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-bottom: 0px;
                            padding:5px"),
                                               wellPanel(withSpinner(DT::dataTableOutput("datatable",
                                                                                         height = "150px")),
                                                         style = "background-color: #ffffff;
                            border-bottom-color: #333333;
                            border-left-color: #333333;
                            border-right-color: #333333;
                            height: 175px;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-top: 0px")
                                        ),
                                        column(width = 8, #style="padding:0px;",
                                               wellPanel("Dimensionality reduction of matrix",
                                                         style = "background-color: #333333;
                            color: white;
                            border-top-color: #333333;
                            border-left-color: #333333;
                            border-right-color: #333333;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-bottom: 0px;
                            padding:5px"),
                                               wellPanel(
                                                 fluidRow(
                                                   column(width = 2,
                                                          radioButtons("method", "Method:",
                                                                       choices = c("t-SNE", "UMAP"))),
                                                   column(width = 2,
                                                          numericInput("perplexity", "Perplexity:",
                                                                       value = 2, min = 1, max = nrow(data) - 1)),
                                                   column(width = 2,
                                                          radioButtons("label", "Labels:",
                                                                       choices = c("Index", "IDs"))),
                                                   column(width = 2,
                                                          numericInput("labelsize", "Label size:",
                                                                       value = 12, min = 1, max = 30)),
                                                   column(width = 8, style = "padding:0px;",
                                                          withSpinner(plotlyOutput("tsne_plot",
                                                                                   height = 550))),
                                                   column(width = 4, style = "padding:0px;",
                                                          withSpinner(plotOutput("tsne_plot_legend",
                                                                                 height = 550))),
                                                   column(width = 2,
                                                          downloadLink("download_plot_data", label = "Download data"))),
                                                 style = "background-color: white;
                            border-bottom-color: #333333;
                            border-left-color:  #333333;
                            border-right-color: #333333;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-top: 0px"
                                                 #height=575px
                                               ))),
                               fluidRow(column(width = 12,
                                               wellPanel("Hierarchical clustering of matrix",
                                                         style = "background-color: #333333;
                            color: white;
                            border-top-color: #333333;
                            border-left-color: #333333;
                            border-right-color: #333333;
                            box-shadow: 3px 3px 3px #d8d8d8;
                            margin-bottom: 0px;
                            padding:5px")
                                               ,
                                               wellPanel(
                                                 fluidRow(
                                                   column(width = 2,
                                                          radioButtons("hcmethod", "Method:",
                                                                       choices = c("ward.D2", "average", "complete", "single"))),
                                                   column(width = 2,
                                                          numericInput("labelsize_hc", "Label size:",
                                                                       value = 8, min = 1, max = 30))
                                                 ),
                                                 fluidRow(
                                                   column(width = 9,
                                                          withSpinner(plotOutput("hclust"))),
                                                   column(width = 3,
                                                          withSpinner(plotOutput("hclust_legend"))),
                                                   column(width = 10,
                                                          h4("Adjusted rand index")),
                                                   column(width = 4,
                                                          numericInput("kclusters", "Select number of clusters to compute adjusted rand index with:",
                                                                       value = length(unique(as.character(data[, index_grouping[1]]))), min = 2, max = nrow(data))),
                                                   column(width = 10,
                                                          br(),
                                                          uiOutput("hclust_ari"),
                                                          HTML("<p><a href='https://link.springer.com/article/10.1007%2FBF01908075' target='_blank'>The adjusted rand index</a> compares two partitions of the same data and ranges from -1 to 1, with 0 showing random labeling and 1 showing perfect agreement between the two partitions.</p>"))
                                                 ),
                                                 style = "background-color: #ffffff;
                                border-bottom-color: #333333;
                                border-left-color: #333333;
                                border-right-color: #333333;
                                box-shadow: 3px 3px 3px #d8d8d8;
                                margin-top: 0px")
                                               ,
                                               verbatimTextOutput("test")
                               ))
                      ))
  )))

###### SERVER ######
server <- function(input, output, session) {

##### Global #####
  ids <- reactive({
    paste0(data[[colindex_id]], " (", seq(1, length(data[[colindex_id]])), ")")
  })
  index_id <- reactive({
    which(ids() == input$id_interest)
  })

  ##### Wordcloud plot and download#####
  output$id_text <- renderText({
    paste("Wordcloud of", data[[colindex_id]][index_id()])
  })

  output$wordcloud_plot <- renderPlot({
    id_matrix <- matrix[index_id(), ]
    id_matrix <- data.frame(word = as.character(names(id_matrix)),
                            freq = id_matrix)
    colnames(id_matrix) <- c("word", "freq")
    id_matrix <- id_matrix[id_matrix$freq == 1, ]
    plotWordCloud(id_matrix,
                  max.words = min(nrow(id_matrix), input$nword),
                  scale = c(input$fontsize / 10, input$fontsize / 10),
                  colors = brewer.pal(8, "Greys")[4:8])
  })

  output$download_wordcloud <- downloadHandler(
    filename <- function() {
      paste0(paste0("Wordcloudof",
                    data[[colindex_id]][index_id()]), ".pdf", sep = "")
    },
    content <- function(file) {
      id_matrix <- matrix[index_id(), ]
      id_matrix <- data.frame(word = names(id_matrix), freq = id_matrix)
      colnames(id_matrix) <- c("word", "freq")
      id_matrix <- id_matrix[id_matrix$freq == 1, ]
      pdf(file)
      plotWordCloud(id_matrix,
                    max.words = min(nrow(id_matrix), input$nword),
                    scale = c(input$fontsize / 10, input$fontsize / 10),
                    colors = brewer.pal(8, "Greys")[4:8])
      dev.off()
    }
  )

  ##### Table #####
  output$table  <-  renderText({
    paste("Most occuring words among IDs")
  })

  output$datatable <- DT::renderDataTable({
    colsum_data <- data.frame(word = colnames(matrix), freq = colSums(matrix))
    colsum_data <- colsum_data[order(colsum_data$freq, decreasing = T), ]
    colnames(colsum_data) <- c("Word", paste0("IDs (total n=", nrow(matrix), ")"))
    DT::datatable(colsum_data,
                  extensions = c("Buttons"),
                  rownames = F,
                  fillContainer = T,
                  escape = FALSE,
                  options = list(dom = "t",
                                 scrollY = min(nrow(colsum_data), 500),
                                 scrollX = TRUE,
                                 scroller = TRUE,
                                 autoWidth = TRUE,
                                 pageLength = nrow(colsum_data),
                                 columnDefs = list(
                                   list(className = "dt-center", targets = "_all"),
                                   list(width = "50%", targets = "_all")))
    )
  })

  ##### colour #####
  outvar <- reactive({
    if (input$colour == "Grouping variable") {
      return(names(data)[index_grouping])
    } else {
      return(colnames(matrix))
    }
  })

  observe({
    updateSelectInput(session, "colour_select", choices = outvar())})

  colour_choice <- reactive({
    if (input$colour == "Grouping variable") {
      return(as.factor(data[, input$colour_select]))
    } else {
      matrix <- as.data.frame(matrix)
      colour_byword <- matrix[[input$colour_select]]
      colour_byword <- ifelse(colour_byword > 0,
                              "Selected word associated with ID",
                              "Selected word not associated with ID")
      return(as.factor(colour_byword))
    }
  })

  color_palette <- reactive({
    palette <- c("#A6CEE3", "#1F78B4", "#B2DF8A", "#33A02C", "#FB9A99",
                 "#E31A1C", "#FDBF6F", "#FF7F00", "#CAB2D6", "#6A3D9A",
                 "#00AFBB", "#E7B800", "#FC4E07", "#999999", "#E69F00",
                 "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#D55E00")
    return(palette[seq(length(levels(colour_choice())))])
  })

  ##### Dimension reduction plot and download #####
  data_dimred <- reactive({
    if (input$method == "t-SNE") {
      tsne_result <- Rtsne(matrix, perplexity = input$perplexity, check_duplicates = F)
      data["X_Coord"] <- tsne_result$Y[, 1]
      data["Y_Coord"] <- tsne_result$Y[, 2]
      return(data)
    } else if (input$method == "UMAP") {
      umap_result <- umap(matrix)
      data["X_Coord"] <- umap_result$layout[, 1]
      data["Y_Coord"] <- umap_result$layout[, 2]
      return(data)
    }
  })

  output$tsne_plot <- renderPlotly({
    if (input$label == "Index") {
      labeling <- as.character(seq(1, nrow(data)))
    } else if (input$label == "IDs") {
      labeling <- as.character(data[[colindex_id]])
    }
    p <- plot_ly(colors = color_palette()) %>%
      add_trace(type = "scatter",
                mode = "markers",
                x = data_dimred()$X_Coord[index_id()],
                y = data_dimred()$Y_Coord[index_id()],
                opacity = 0.15,
                marker = list(
                  color = "grey",
                  size = 80)) %>%
      add_trace(x = data_dimred()$X_Coord,
                y = data_dimred()$Y_Coord,
                type = "scatter",
                mode = "text",
                text = labeling,
                textfont = list(size = input$labelsize),
                color = factor(colour_choice())) %>%
      add_trace(x = data_dimred()$X_Coord,
                y = data_dimred()$Y_Coord,
                type = "scatter",
                mode = "markers",
                opacity = 0,
                text = paste0("ID: ", data[[colindex_id]], "\n",
                              "Index: ", seq(1, nrow(data)), "\n",
                              "Grouping: ", paste(data[, index_grouping])),
                hoverinfo = "text",
                color = factor(colour_choice())) %>%
      layout(showlegend = FALSE,
             yaxis = list(title = "",
                          zeroline = FALSE,
                          linecolor = toRGB("black"),
                          linewidth = 1,
                          showticklabels = FALSE,
                          showgrid = FALSE),
             xaxis = list(title = "",
                          zeroline = FALSE,
                          linecolor = toRGB("black"),
                          linewidth = 1,
                          showticklabels = FALSE,
                          showgrid = FALSE),
             autosize = T) %>%
      config(modeBarButtonsToRemove = c("zoomIn2d", "zoomOut2d",
                                        "hoverClosestGeo", "hoverClosestGl2d",
                                        "toImage", "hoverClosestCartesian",
                                        "lasso2d", "select2d", "resetScale2d",
                                        "hoverCompareCartesian",
                                        "hoverClosestPie", "toggleSpikelines"),
             displaylogo = FALSE) %>%
      style(hoverinfo = "none", traces = c(1, 2))
    p
  })

  #legend of plotly plot by ggplot
  output$tsne_plot_legend <- renderPlot({
    p <- ggplot(data, aes(x = 1, y = 1)) +
      geom_text(aes(label = seq(1, nrow(data)), colour = factor(colour_choice())),
                size = 3.5, fontface = "bold") +
      theme_classic() +
      scale_color_manual(values = color_palette()) +
      theme(legend.title = element_blank()) +
      theme(legend.position = "right") +
      theme(legend.text = element_text(size = 9))
    leg <- get_legend(p)
    as_ggplot(leg)
  })

  output$download_plot_data <- downloadHandler(
    filename <- function() {
      paste0(input$method, "_coordinates.csv")
    },
    content <- function(file) {
      write.csv(data_dimred(), file, row.names = F)
    }
  )

  ##### Hierarchical clustering #######
  output$hclust <- renderPlot({
    set.seed(42)
    clustering <- hclust(dist(matrix), method = input$hcmethod)
    par(oma = c(3, 3, 3, 3))
    palette(color_palette())
    par(mar = rep(0, 4))
    myplclust(clustering,
              labels = paste(data[[colindex_id]]),
              lab.col = as.fumeric(as.character(colour_choice()),
                                   levels = sort(unique(as.character(colour_choice())))),
              cex = as.numeric(input$labelsize_hc / 10),
              main = "",
              yaxt = "n",
              ylab = "")
  })

  #legend
  output$hclust_legend <- renderPlot({
    p <- ggplot(data, aes(x = 1, y = 1)) +
      geom_text(aes(label = seq(1, nrow(data)),
                    colour = factor(colour_choice())),
                fontface = "bold") +
      theme_classic() +
      scale_color_manual(values = color_palette()) +
      theme(legend.title = element_blank()) +
      theme(legend.position = "right") +
      theme(legend.text = element_text(size = 9))
    leg <- get_legend(p)
    as_ggplot(leg)
  })

  output$hclust_ari <- renderUI({
    #adjusted Rand index
    clustering <- hclust(dist(matrix), method = input$hcmethod)
    mc_ari <- adjustedRandIndex(as.character(data[, index_grouping[1]]), cutree(clustering, k = input$kclusters))
    text1 <- paste("By dividing the data into", input$kclusters, "clusters and comparing the resulting partition with the initial grouping variable, the adjusted rand index equals")
    text2 <- paste(round(mc_ari, 2), ".")
    HTML(paste(text1, text2))
  })
  ##### Test field for development ######
  #output$test <- renderPrint({
  #})
}

###### APP ######
shinyApp(ui, server)