server.R

# server.R
library(shiny)
library(readxl)
library(DT)
library(dplyr)

# Source the warning overlay and long operations code
source("warning_overlay.R", local = TRUE)
source("long_operations.R", local = TRUE)
source("auth/hf_oauth.R", local = TRUE)
source("utils/supabase_r.R", local = TRUE)

# setwd("/Users/audrey/Downloads/research/ckweb/Tcellstates")

# Define server logic
function(input, output, session) {

  # --- START: Environment Variable Diagnostic ---
  print("=== Environment Variables Diagnostic ===")
  all_env <- Sys.getenv()
  # Print only relevant environment variables (filter sensitive ones)
  relevant_vars <- c("SUPABASE_URL", "SUPABASE_KEY", "OAUTH_CLIENT_ID", "OAUTH_CLIENT_SECRET",
                     "SPACE_HOST", "SPACE_ID", "HF_TOKEN")
  for (var_name in relevant_vars) {
    var_value <- Sys.getenv(var_name, unset = "NOT_SET")
    if (var_name %in% c("SUPABASE_KEY", "OAUTH_CLIENT_SECRET", "HF_TOKEN")) {
      # Mask sensitive values
      print(paste(var_name, ":", if(var_value == "NOT_SET") "NOT_SET" else "[REDACTED]"))
    } else {
      print(paste(var_name, ":", var_value))
    }
  }
  print("========================================")
  # --- END: Environment Variable Diagnostic ---

  # --- START: OAuth and Supabase Initialization ---
  oauth_config <- initialize_oauth()
  supabase_client <- initialize_supabase()
  # --- END: OAuth and Supabase Initialization ---

  # --- START: TaijiChat R Callback for Python Agent Thoughts ---
  python_agent_thought_callback <- function(thought_message_from_python) {
    # Attempt to explicitly convert to R character and clean up
    thought_message_text <- tryCatch({
      as.character(thought_message_from_python)[1] # Take the first element after converting
    }, error = function(e) {
      print(paste("R Callback: Error converting thought to character:", e$message))
      return(NULL)
    })

    if (!is.null(thought_message_text) && is.character(thought_message_text) && length(thought_message_text) == 1 && nzchar(trimws(thought_message_text))) {
      # print(paste("R Callback: Valid thought from Python -", thought_message_text)) # For R console debugging
      session$sendCustomMessage(type = "agent_new_thought", message = list(text = trimws(thought_message_text)))
    } else {
      # Log the original and potentially converted type for better debugging
      print(paste("R Callback: Received invalid or empty thought. Original type:", class(thought_message_from_python), ", Value:", thought_message_from_python, ", Converted text:", thought_message_text))
    }
  }
  # --- END: TaijiChat R Callback for Python Agent Thoughts ---
  
  # --- START: TaijiChat Agent Initialization ---
  # This assumes manager_agent_module is globally available from ui.R's sourcing.
  # ui.R does: manager_agent_module <- reticulate::import("agents.manager_agent")

  api_key_val <- NULL
  # First try to get API key from environment variable
  api_key_val <- Sys.getenv("OPENAI_API_KEY")
  
  # If environment variable is not set, try reading from file
  if (api_key_val == "") {
    tryCatch({
      api_key_content <- readLines("api_key.txt", warn = FALSE)
      if (length(api_key_content) > 0 && nzchar(trimws(api_key_content[1]))) {
        api_key_val <- trimws(api_key_content[1])
        print("TaijiChat: API key successfully read from file in server.R.")
      } else {
        warning("TaijiChat: api_key.txt is empty or not found. LLM features may be disabled.")
        print("TaijiChat: api_key.txt is empty or not found.")
      }
    }, error = function(e) {
      warning(paste("TaijiChat: Error reading api_key.txt in server.R:", e$message))
      print(paste("TaijiChat: Error reading api_key.txt:", e$message))
    })
  } else {
    print("TaijiChat: API key successfully read from environment variable.")
  }

  py_openai_client_instance <- NULL
  if (!is.null(api_key_val)) {
    tryCatch({
      # Ensure reticulate is configured to use the correct Python environment
      if (reticulate::py_available(initialize = TRUE)) {
        openai_py_module <- reticulate::import("openai", convert = FALSE) # convert=FALSE for raw Python objects
        py_openai_client_instance <- openai_py_module$OpenAI(api_key = api_key_val)
        print("TaijiChat: Python OpenAI client initialized successfully in server.R via reticulate.")
      } else {
        warning("TaijiChat: Python (reticulate) not available or not initialized. Cannot create OpenAI client.")
        print("TaijiChat: Python (reticulate) not available. Cannot create OpenAI client.")
      }
    }, error = function(e) {
      warning(paste("TaijiChat: Failed to initialize Python OpenAI client in server.R:", e$message))
      print(paste("TaijiChat: Failed to initialize Python OpenAI client:", e$message))
      py_openai_client_instance <- NULL 
    })
  } else {
    print("TaijiChat: API key is NULL, skipping Python OpenAI client initialization.")
  }

  rv_agent_instance <- reactiveVal(NULL)

  # Attempt to create the agent instance once.
  current_manager_agent_module <- NULL
  tryCatch({
      # Force reload the module to ensure we get the latest version
      print("TaijiChat: Attempting to reload Python modules to ensure latest version...")
      reticulate::py_run_string("
import sys
import importlib
if 'agents.manager_agent' in sys.modules:
    print('Reloading agents.manager_agent module...')
    importlib.reload(sys.modules['agents.manager_agent'])
")
      
      current_manager_agent_module <- reticulate::import("agents.manager_agent", convert = FALSE)
      if (is.null(current_manager_agent_module)) { 
          warning("TaijiChat: reticulate::import('agents.manager_agent') returned NULL in server.R.")
          print("TaijiChat: reticulate::import('agents.manager_agent') returned NULL in server.R.")
      } else {
          print("TaijiChat: Successfully imported/retrieved 'agents.manager_agent' module in server.R.")
      }
  }, error = function(e) {
      warning(paste("TaijiChat: Failed to import agents.manager_agent in server.R:", e$message))
      print(paste("TaijiChat: Failed to import agents.manager_agent in server.R:", e$message))
  })

  if (!is.null(current_manager_agent_module)) {
      # Module is available, now try to instantiate the agent
      if (!is.null(py_openai_client_instance)) {
          tryCatch({
              supabase_py_client <- if (!is.null(supabase_client)) supabase_client else NULL

              agent_inst <- current_manager_agent_module$ManagerAgent(
                openai_client = py_openai_client_instance,
                r_callback_fn = python_agent_thought_callback,
                supabase_client = supabase_py_client,
                user_id = NULL,
                hf_user_id = NULL
              )
              rv_agent_instance(agent_inst)
              print("TaijiChat: Python ManagerAgent instance created in server.R using pre-initialized client and R callback.")
          }, error = function(e) {
              warning(paste("TaijiChat: Failed to instantiate ManagerAgent in server.R with client & callback:", e$message))
              print(paste("TaijiChat: Failed to instantiate ManagerAgent with client & callback:", e$message))
          })
      } else if (!is.null(api_key_val)) { # Try with API key if client object failed but key exists
          tryCatch({
              supabase_py_client <- if (!is.null(supabase_client)) supabase_client else NULL

              agent_inst <- current_manager_agent_module$ManagerAgent(
                openai_api_key = api_key_val,
                r_callback_fn = python_agent_thought_callback,
                supabase_client = supabase_py_client,
                user_id = NULL,
                hf_user_id = NULL
              )
              rv_agent_instance(agent_inst)
              print("TaijiChat: Python ManagerAgent instance created in server.R with API key and R callback (client to be init by Python).")
          }, error = function(e) {
              warning(paste("TaijiChat: Failed to instantiate ManagerAgent with API key & callback in server.R:", e$message))
              print(paste("TaijiChat: Failed to instantiate ManagerAgent with API key & callback:", e$message))
          })
      } else {
          # Neither client nor API key is available for the agent
          warning("TaijiChat: Cannot create ManagerAgent instance: OpenAI client/API key not available for agent constructor.")
          print("TaijiChat: Cannot create ManagerAgent: OpenAI client/API key not available for agent constructor.")
      }
  } else {
      # Module itself could not be imported/retrieved
      warning("TaijiChat: agents.manager_agent module is NULL after import attempt. Agent not created.")
      print("TaijiChat: agents.manager_agent module is NULL after import attempt. Agent not created.")
  }
  # --- END: TaijiChat Agent Initialization ---

  # --- START: Authentication Handlers ---

  # Handle login button click - generate OAuth URL (only once)
  observeEvent(input$hfSignInBtn, {
    req(input$hfSignInBtn)

    # Check if we're in OAuth callback (code in URL)
    query_string <- parseQueryString(session$clientData$url_search)
    if (!is.null(query_string$code)) {
      print("OAuth: In callback, skipping button redirect")
      return()
    }

    if (!oauth_config$enabled) {
      print("OAuth: Not enabled")
      return()
    }

    print(paste("OAuth: Button clicked, count:", input$hfSignInBtn))

    # Generate random state for CSRF protection
    state <- generate_oauth_state()
    session$userData$oauth_state <- state

    # Get base URL - use SPACE_HOST if available (HF Spaces), otherwise use client URL
    space_host <- Sys.getenv("SPACE_HOST", "")
    if (space_host != "") {
      redirect_uri <- paste0("https://", space_host)
    } else {
      redirect_uri <- paste0(session$clientData$url_protocol, "//",
                            session$clientData$url_hostname,
                            if (session$clientData$url_port != "") paste0(":", session$clientData$url_port) else "")
    }

    print(paste("OAuth: Redirect URI:", redirect_uri))

    # Generate authorization URL
    auth_url <- get_authorization_url(oauth_config, redirect_uri, state)

    if (!is.null(auth_url)) {
      print(paste("OAuth: Redirecting to:", auth_url))
      # Send redirect message to JavaScript
      session$sendCustomMessage(type = 'redirect_to_oauth', message = list(url = auth_url))
    }
  }, ignoreInit = TRUE, once = TRUE)

  # Send initial auth state on session start
  # Don't send if we're in OAuth callback (will be handled after auth completes)
  observe({
    isolate({
      # Check if we're in OAuth callback by looking at URL parameters
      query_string <- parseQueryString(session$clientData$url_search)
      if (is.null(query_string$code)) {
        # No OAuth code in URL, show login overlay
        session$sendCustomMessage('auth_state', list(authenticated = FALSE))
      }
    })
  })

  # OAuth callback handler
  observeEvent(input$oauth_code, {
    if (!oauth_config$enabled || is.null(supabase_client)) {
      print("OAuth: Not enabled or Supabase not configured")
      return()
    }

    tryCatch({
      # Get redirect URI - must match exactly what was used in authorization request
      space_host <- Sys.getenv("SPACE_HOST", "")
      if (space_host != "") {
        redirect_uri <- paste0("https://", space_host)
      } else {
        redirect_uri <- paste0(session$clientData$url_protocol, "//",
                              session$clientData$url_hostname,
                              if (session$clientData$url_port != "") paste0(":", session$clientData$url_port) else "")
      }

      print(paste("OAuth: Token exchange using redirect_uri:", redirect_uri))

      # Exchange code for token
      token_result <- exchange_code_for_token(oauth_config, input$oauth_code, redirect_uri)

      if (is.null(token_result)) {
        print("OAuth: Failed to exchange code for token")
        return()
      }

      # Get user info from HF
      user_info <- get_user_info(token_result$access_token)

      if (is.null(user_info)) {
        print("OAuth: Failed to get user info")
        return()
      }

      # Create or retrieve user in Supabase
      supabase_user <- get_or_create_user(
        supabase_client,
        user_info$hf_user_id,
        user_info$hf_username,
        user_info$email
      )

      if (is.null(supabase_user)) {
        print("OAuth: Failed to create/retrieve user in Supabase")
        return()
      }

      # Store in session
      session$userData$hf_user <- user_info
      session$userData$supabase_user <- supabase_user
      session$userData$access_token <- token_result$access_token

      print(paste("OAuth: User authenticated -", user_info$hf_username))

      # Hide login overlay
      session$sendCustomMessage('auth_state', list(authenticated = TRUE))

      # Show user info with quota
      tokens_used <- supabase_user$tokens_used %||% 0
      token_quota <- supabase_user$token_quota %||% 100000
      quota_text <- paste(tokens_used, "/", token_quota)

      session$sendCustomMessage('update_user_info', list(
        username = user_info$hf_username,
        quota = quota_text
      ))

    }, error = function(e) {
      print(paste("OAuth: Error in callback handler -", e$message))
    })
  })

  # Logout handler
  observeEvent(input$logoutBtn, {
    logout_user(session)
    session$sendCustomMessage('auth_state', list(authenticated = FALSE))
    print("OAuth: User logged out")
  })

  # --- END: Authentication Handlers ---

  # Server logic for home tab
  output$home <- renderText({
    "Welcome to the Home page"
  })
  
  observeEvent(input$read_now, {
    # Trigger a redirect using JavaScript
    session$sendCustomMessage(type = "redirect", 
                              message = "https://doi.org/10.1101/2023.01.03.522354")
  })
  
  
  # NEW READ EXCEL FILE FOR TRANSPOSED DATASET
  new_read_excel_file <- function(path) {
    df <- read_excel(path)
    colnames(df)[1] <- "Regulator Names"
    
    # Transpose the dataframe
    df_transposed <- as.data.frame(t(df))
    
    # Fix the column names of the transposed dataframe (optional)
    colnames(df_transposed) <- df_transposed[1, ]  # Set first row as column names
    df_transposed <- df_transposed[-1, ]  # Remove the first row which is now used as column names
    
    return(df_transposed)
  }
  
  # # NEW FILTER FUNCTION FOR TRANSPOSED DATASET
  # new_filter_data <- function(df, keyword) {
  # 
  #   # Find the columns whose names contain the keyword
  #   matching_columns <- grepl(keyword, colnames(df), ignore.case = TRUE)
  # 
  #   # Check if any matching columns exist
  #   if (sum(matching_columns) == 0) {
  #     # If no matching columns are found, return an empty dataframe with proper structure
  #     return(data.frame())
  #   }
  # 
  #   # Subset the dataframe, ensuring the result is always a dataframe
  #   filtered_df <- df[, matching_columns, drop = FALSE]
  # 
  #   return(filtered_df)
  # }
  
  #NEW FILTER FUNCTION FOR MULTIPLE GENE SEARCH
  # new_filter_data <- function(df, keywords) {
  #   # Split the keywords by commas and remove any leading/trailing whitespace
  #   keyword_list <- strsplit(keywords, ",")[[1]]
  #   keyword_list <- trimws(keyword_list)
  # 
  #   # Initialize an empty logical vector for matching columns
  #   matching_columns <- rep(FALSE, ncol(df))
  # 
  #   # Loop through each keyword and update the matching_columns vector
  #   for (keyword in keyword_list) {
  #     matching_columns <- matching_columns | grepl(keyword, colnames(df), ignore.case = TRUE)
  #   }
  # 
  #   # Check if any matching columns exist
  #   if (sum(matching_columns) == 0) {
  #     # If no matching columns are found, return an empty dataframe with proper structure
  #     return(data.frame())
  #   }
  # 
  #   # Subset the dataframe, ensuring the result is always a dataframe
  #   filtered_df <- df[, matching_columns, drop = FALSE]
  # 
  #   return(filtered_df)
  # }
  
  new_filter_data <- function(df, keywords) {
    # If no keywords are provided, return the full dataset
    if (is.null(keywords) || keywords == "") {
      return(df)
    }
    
    # Split the keywords by commas and remove any leading/trailing whitespace
    keyword_list <- strsplit(keywords, ",")[[1]]
    keyword_list <- trimws(keyword_list)  # Remove leading/trailing spaces
    
    # Initialize an empty logical vector for matching columns
    matching_columns <- rep(FALSE, ncol(df))
    
    # Loop through each keyword and update the matching_columns vector
    for (keyword in keyword_list) {
      matching_columns <- matching_columns | grepl(keyword, colnames(df), ignore.case = TRUE)
    }
    
    # Check if any matching columns exist
    if (sum(matching_columns) == 0) {
      # If no matching columns are found, return an empty dataframe
      return(data.frame())
    }
    
    # Subset the dataframe, ensuring the result is always a dataframe
    filtered_df <- df[, matching_columns, drop = FALSE]
    
    return(filtered_df)
  }
  
  

  
  # TESTING FUNCTIONS
  
  #NEW ALL DATA SEARCH
  data <- reactive({
    new_read_excel_file("www/tablePagerank/Table_TF PageRank Scores for Audrey.xlsx")
  })
  
  # Track the current column page
  column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for main page data
  observeEvent(input$search_input, {
    column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked
  observeEvent(input$next_btn, {
    current_page <- column_page()
    total_cols <- ncol(new_filter_data(data(), input$search_input))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      column_page(current_page + 1)
    }
    
    # new_page <- column_page() + 1
    # column_page(new_page)
  })
  
  observeEvent(input$prev_btn, {
    current_page <- column_page()
    column_page(max(current_page - 1, 1))
    # new_page <- max(column_page() - 1, 1)  # Ensure the page does not go below 1
    # column_page(new_page)
  })
  
  # # OLD: Reactive for filtering the data based on input$search_input
  # new_filtered_data <- reactive({
  # 
  #   df <- new_filter_data(data(), input$search_input)
  # 
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  # 
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  # 
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  # 
  #   # Ensure start_col is within the number of columns
  #   if (start_col > total_cols) {
  #     return(df)  # Return the filtered dataframe as-is if start_col exceeds the number of columns
  #   }
  # 
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   return(df_subset)
  # })
  
  #NEW FILTER DATA TO INCLUDE ADDITIONAL DESCRIPTION ROW + WORKS WITH SEARCHING INDIVIDUAL GENES
  new_filtered_data <- reactive({
    # Filter the data and determine the total number of columns
    df <- new_filter_data(data(), input$search_input)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold)
      style <- "<h5 style='font-weight: bold;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Combine the data with the description and styled "Cell state data" row
      df_subset <- rbind(df_subset[1:2, , drop = FALSE], description_row, df_subset[-(1:2), , drop = FALSE])
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 2;  // Adjust this index to the row you want highlighted

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })

  # Rendering the filtered data table and showing all rows
  output$table <- renderDT({
    new_filtered_data()
  }, options = list(pageLength = nrow(new_filtered_data()), dom = 't'))
  
  
  
  
  #NEW NAIVE SEARCH
  naive_data <- reactive({
    new_read_excel_file("www/tablePagerank/Naive.xlsx")
  })
  
  # Track the current column page for naive data
  naive_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for naive data
  observeEvent(input$search_input_naive, {
    naive_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Reset the column page when the search input changes for naive data
  observeEvent(input$search_input_naive, {
    naive_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for naive data
  observeEvent(input$naive_next_btn, {
    current_page <- naive_column_page()
    total_cols <- ncol(new_filter_data(naive_data(), input$search_input_naive))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      naive_column_page(current_page + 1)
    }
    
    # new_page <- naive_column_page() + 1
    # naive_column_page(new_page)
  })
  
  observeEvent(input$naive_prev_btn, {
    current_page <- naive_column_page()
    naive_column_page(max(current_page - 1, 1)) 
    # new_page <- max(naive_column_page() - 1, 1)  # Ensure the page does not go below 1
    # naive_column_page(new_page)
  })

  #OLD FILTERED DATA
  # filtered_naive_data <- reactive({
  # 
  #   df <- new_filter_data(naive_data(), input$search_input_naive)
  # 
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  # 
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  # 
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (naive_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  # 
  #   # Ensure start_col is within the number of columns
  #   if (start_col > total_cols) {
  #     return(df)  # Return the filtered dataframe as-is if start_col exceeds the number of columns
  #   }
  # 
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_naive_data <- reactive({
    # Filter the naive data and determine the total number of columns
    df <- new_filter_data(naive_data(), input$search_input_naive)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (naive_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })

  output$table_naive <- renderDT({
    filtered_naive_data()
  }, options = list(pageLength = nrow(filtered_naive_data()), dom = 't'))

  

  #NEW TE SEARCH
  te_data <- reactive({
    new_read_excel_file("www/tablePagerank/TE.xlsx")
  })

  # Track the current column page for "te" data
  te_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "te" data
  observeEvent(input$search_input_te, {
    te_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "te" data
  observeEvent(input$te_next_btn, {
    current_page <- te_column_page()
    total_cols <- ncol(new_filter_data(te_data(), input$search_input_te))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      te_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$te_prev_btn, {
    current_page <- te_column_page()
    te_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "te" page data
  # filtered_te_data <- reactive({
  #   df <- new_filter_data(te_data(), input$search_input_te)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (te_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  #   
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_te_data <- reactive({
    # Filter the TE data and determine the total number of columns
    df <- new_filter_data(te_data(), input$search_input_te)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (te_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })
  
  # Rendering the filtered "te" page data table and showing all rows
  output$table_te <- renderDT({
    filtered_te_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_te_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))



  #NEW MP SEARCH
  mp_data <- reactive({
    new_read_excel_file("www/tablePagerank/MP.xlsx")
  })

  # Track the current column page for "mp" data
  mp_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "mp" data
  observeEvent(input$search_input_mp, {
    mp_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "mp" data
  observeEvent(input$mp_next_btn, {
    current_page <- mp_column_page()
    total_cols <- ncol(new_filter_data(mp_data(), input$search_input_mp))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      mp_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$mp_prev_btn, {
    current_page <- mp_column_page()
    mp_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "mp" page data
  # filtered_mp_data <- reactive({
  #   df <- new_filter_data(mp_data(), input$search_input_mp)
  # 
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  # 
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  # 
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (mp_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  # 
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  # 
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_mp_data <- reactive({
    # Filter the MP data and determine the total number of columns
    df <- new_filter_data(mp_data(), input$search_input_mp)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (mp_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })

  # Rendering the filtered "mp" page data table and showing all rows
  output$table_mp <- renderDT({
    filtered_mp_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_mp_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))

  

  #NEW TCM SEARCH
  tcm_data <- reactive({
    new_read_excel_file("www/tablePagerank/TCM.xlsx")
  })

  # Track the current column page for "tcm" data
  tcm_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "tcm" data
  observeEvent(input$search_input_tcm, {
    tcm_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "tcm" data
  observeEvent(input$tcm_next_btn, {
    current_page <- tcm_column_page()
    total_cols <- ncol(new_filter_data(tcm_data(), input$search_input_tcm))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      tcm_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$tcm_prev_btn, {
    current_page <- tcm_column_page()
    tcm_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "tcm" page data
  # filtered_tcm_data <- reactive({
  #   df <- new_filter_data(tcm_data(), input$search_input_tcm)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (tcm_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  # 
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  # 
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_tcm_data <- reactive({
    # Filter the TCM data and determine the total number of columns
    df <- new_filter_data(tcm_data(), input$search_input_tcm)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (tcm_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })
  
  # Rendering the filtered "tcm" page data table and showing all rows
  output$table_tcm <- renderDT({
    filtered_tcm_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_tcm_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))


  #NEW TEM SEARCH
  tem_data <- reactive({
    new_read_excel_file("www/tablePagerank/TEM.xlsx")
  })

  # Track the current column page for "tem" data
  tem_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "tem" data
  observeEvent(input$search_input_tem, {
    tem_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "tem" data
  observeEvent(input$tem_next_btn, {
    current_page <- tem_column_page()
    total_cols <- ncol(new_filter_data(tem_data(), input$search_input_tem))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      tem_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$tem_prev_btn, {
    current_page <- tem_column_page()
    tem_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "tem" page data
  # filtered_tem_data <- reactive({
  #   df <- new_filter_data(tem_data(), input$search_input_tem)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (tem_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  #   
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_tem_data <- reactive({
    # Filter the TEM data and determine the total number of columns
    df <- new_filter_data(tem_data(), input$search_input_tem)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (tem_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })
  
  # Rendering the filtered "tem" page data table and showing all rows
  output$table_tem <- renderDT({
    filtered_tem_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_tem_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))

  #NEW TRM SEARCH
  trm_data <- reactive({
    new_read_excel_file("www/tablePagerank/TRM.xlsx")
  })

  # Track the current column page for "trmm" data
  trm_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "trmm" data
  observeEvent(input$search_input_trm, {
    trm_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "trmm" data
  observeEvent(input$trm_next_btn, {
    current_page <- trm_column_page()
    total_cols <- ncol(new_filter_data(trm_data(), input$search_input_trm))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      trm_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$trm_prev_btn, {
    current_page <- trm_column_page()
    trm_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "trmm" page data
  # filtered_trm_data <- reactive({
  #   df <- new_filter_data(trm_data(), input$search_input_trm)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (trm_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  #   
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_trm_data <- reactive({
    # Filter the TRM data and determine the total number of columns
    df <- new_filter_data(trm_data(), input$search_input_trm)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (trm_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })
  
  # Rendering the filtered "trmm" page data table and showing all rows
  output$table_trm <- renderDT({
    filtered_trm_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_trm_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))


  #NEW TEX PROG SEARCH
  texprog_data <- reactive({
    new_read_excel_file("www/tablePagerank/TEXprog.xlsx")
  })

  # Track the current column page for "texprog" data
  texprog_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "texprog" data
  observeEvent(input$search_input_texprog, {
    texprog_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "texprog" data
  observeEvent(input$texprog_next_btn, {
    current_page <- texprog_column_page()
    total_cols <- ncol(new_filter_data(texprog_data(), input$search_input_texprog))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      texprog_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$texprog_prev_btn, {
    current_page <- texprog_column_page()
    texprog_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "texprog" page data
  # filtered_texprog_data <- reactive({
  #   df <- new_filter_data(texprog_data(), input$search_input_texprog)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (texprog_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  #   
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_texprog_data <- reactive({
    # Filter the TEXprog data and determine the total number of columns
    df <- new_filter_data(texprog_data(), input$search_input_texprog)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (texprog_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })
  
  # Rendering the filtered "texprog" page data table and showing all rows
  output$table_texprog <- renderDT({
    filtered_texprog_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_texprog_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))


  #NEW TEX EFF LIKE SEARCH
  texefflike_data <- reactive({
    new_read_excel_file("www/tablePagerank/TEXeff.xlsx")
  })

  # Track the current column page for "texefflike" data
  texefflike_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "texefflike" data
  observeEvent(input$search_input_texefflike, {
    texefflike_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "texefflike" data
  observeEvent(input$texefflike_next_btn, {
    current_page <- texefflike_column_page()
    total_cols <- ncol(new_filter_data(texefflike_data(), input$search_input_texefflike))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      texefflike_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$texefflike_prev_btn, {
    current_page <- texefflike_column_page()
    texefflike_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "texefflike" page data
  # filtered_texefflike_data <- reactive({
  #   df <- new_filter_data(texefflike_data(), input$search_input_texefflike)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (texefflike_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  #   
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_texefflike_data <- reactive({
    # Filter the TEXefflike data and determine the total number of columns
    df <- new_filter_data(texefflike_data(), input$search_input_texefflike)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (texefflike_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })
  
  # Rendering the filtered "texefflike" page data table and showing all rows
  output$table_texefflike <- renderDT({
    filtered_texefflike_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_texefflike_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))


  
  #NEW TEX TERM SEARCH
  texterm_data <- reactive({
    new_read_excel_file("www/tablePagerank/TEXterm.xlsx")
  })

  # Track the current column page for "texterm" data
  texterm_column_page <- reactiveVal(1)
  
  # Reset the column page when the search input changes for "texterm" data
  observeEvent(input$search_input_texterm, {
    texterm_column_page(1)  # Reset to page 1 when a new search is made
  })
  
  # Update the column page when buttons are clicked for "texterm" data
  observeEvent(input$texterm_next_btn, {
    current_page <- texterm_column_page()
    total_cols <- ncol(new_filter_data(texterm_data(), input$search_input_texterm))
    
    # Calculate the maximum possible number of pages
    max_page <- ceiling(total_cols / 4)
    
    # Move to the next page, but do not exceed the max_page
    if (current_page < max_page) {
      texterm_column_page(current_page + 1)
    }
  })
  
  observeEvent(input$texterm_prev_btn, {
    current_page <- texterm_column_page()
    texterm_column_page(max(current_page - 1, 1))  # Ensure the page does not go below 1
  })
  
  #OLD: Reactive for filtering and paginating the "texterm" page data
  # filtered_texterm_data <- reactive({
  #   df <- new_filter_data(texterm_data(), input$search_input_texterm)
  #   
  #   # Get the total number of columns in the filtered dataframe
  #   total_cols <- ncol(df)
  #   
  #   # Ensure there are columns after filtering
  #   if (total_cols == 0) {
  #     return(data.frame())  # Return an empty dataframe if no columns match the search
  #   }
  #   
  #   # Define the start and end index for columns based on the current page
  #   start_col <- (texterm_column_page() - 1) * 4 + 1
  #   end_col <- min(start_col + 3, total_cols)  # Show up to 4 columns
  #   
  #   # If start_col exceeds the total number of columns, return the last valid subset
  #   if (start_col > total_cols) {
  #     start_col <- (ceiling(total_cols / 4) - 1) * 4 + 1  # Set start_col to the last valid page
  #     end_col <- total_cols  # End with the last column
  #   }
  #   
  #   # Subset the columns for the current page
  #   df_subset <- df[, start_col:end_col, drop = FALSE]
  #   
  #   return(df_subset)
  # })
  
  #NEW FILTERED DATA
  filtered_texterm_data <- reactive({
    # Filter the TEXterm data and determine the total number of columns
    df <- new_filter_data(texterm_data(), input$search_input_texterm)
    total_cols <- ncol(df)
    
    # Return empty dataframe if no columns match the search
    if (total_cols == 0) return(data.frame())
    
    # Define the column range for the current page
    start_col <- (texterm_column_page() - 1) * 4 + 1
    end_col <- min(start_col + 3, total_cols)
    
    # Return filtered data if start_col exceeds the total columns
    if (start_col > total_cols) return(df)
    
    # Subset the dataframe for the current page
    df_subset <- df[, start_col:end_col, drop = FALSE]
    
    # Apply HTML formatting to description and "Cell state data" rows, moving them to the first row
    if (nrow(df_subset) >= 2) {
      description_row <- data.frame(
        matrix(NA, nrow = 1, ncol = ncol(df_subset), dimnames = list(NULL, colnames(df_subset)))
      )
      
      # Define the HTML style for the description row (making "TF activity score" bold with white font and background color)
      style <- "<h5 style='font-weight: bold; color: white;'>TF activity score</h5>"
      
      # Apply style to both description and "Cell state data"
      description_row[1, ] <- style
      rownames(description_row)[1] <- "Cell state data"
      
      # Move the description row to the first row of the subset
      df_subset <- rbind(description_row, df_subset)  # Move this styled row to the first position
    }
    
    # Return the data with HTML rendering enabled, show all 45 rows, and remove the search box
    DT::datatable(df_subset, escape = FALSE, options = list(
      pageLength = 45,
      lengthChange = FALSE,
      searching = FALSE,  # Remove the search box
      rowCallback = JS(
        "function(row, data, index) {
        var highlightedRow = 0;  // Apply styling to the first row

        if (index === highlightedRow) {
          $('td', row).css({
            'background-color': '#95a5a6',
            'color': 'white',
            'font-weight': 'bold'
          });
        }
      }"
      )
    ))
  })

  # Rendering the filtered "texterm" page data table and showing all rows
  output$table_texterm <- renderDT({
    filtered_texterm_data()  # Use the filtered data with pagination logic applied
  }, options = list(
    pageLength = nrow(filtered_texterm_data()),  # Show all rows
    dom = 't',  # Remove the row length dropdown
    scrollX = TRUE  # Enable horizontal scrolling if needed
  ))

  
  
  
  # # ORIGINAL READ EXCEL FILE
  # read_excel_file <- function(path) {
  #   df <- read_excel(path)
  #   colnames(df)[1] <- "Regulator Names"
  #   return(df)
  # }
  # 
  # # ORIGINAL FILTER FUNCTION
  # filter_data <- function(data, keyword) {
  #   if (is.null(keyword) || is.na(keyword) || keyword == "") {
  #     return(data)
  #   } else {
  #     keyword <- tolower(keyword)
  #     return(data[apply(data, 1, function(x) any(grepl(keyword, tolower(x)))), ])
  #   }
  # }
  # 
  # #ORIGINAL DISPLAY TABLE FUNCTION
  # # Main page data
  # data <- reactive({
  #   read_excel_file("www/tablePagerank/Table_TF PageRank Scores for Audrey.xlsx")
  # })
  # 
  # filtered_data <- reactive({
  #   filter_data(data(), input$search_input)
  # })
  # 
  # output$table <- renderDT({
  #   filtered_data()
  # })


  
  # Server logic for naive tab
  output$naive <- renderText({
    "TF Activity Score: Naive"
  })
  
  # OLD NAIVE DATA TABLE
  # naive_data <- reactive({
  #   read_excel_file("www/tablePagerank/Naive.xlsx")
  # })
  # 
  # filtered_naive_data <- reactive({
  #   filter_data(naive_data(), input$search_input_naive)
  # })
  # 
  # output$table_naive <- renderDT({
  #   filtered_naive_data()
  # })
  

  # Server logic for TE tab
  output$te <- renderText({
    "TF Activity Score: TE"
  })
  
  # OLD TE DATA TABLE
  # te_data <- reactive({
  #   read_excel_file("www/tablePagerank/TE.xlsx")
  # })
  # 
  # filtered_te_data <- reactive({
  #   filter_data(te_data(), input$search_input_te)
  # })
  # 
  # output$table_te <- renderDT({
  #   filtered_te_data()
  # })

  # Server logic for MP tab
  output$mp <- renderText({
    "TF Activity Score: MP"
  })
  
  # OLD MP DATA
  # mp_data <- reactive({
  #   read_excel_file("www/tablePagerank/MP.xlsx")
  # })
  # 
  # filtered_mp_data <- reactive({
  #   filter_data(mp_data(), input$search_input_mp)
  # })
  # 
  # output$table_mp <- renderDT({
  #   filtered_mp_data()
  # })

  # Server logic for T CM tab
  output$tcm <- renderText({
    "TF Activity Score: T CM"
  })
  
  # OLD TCM DATA
  # tcm_data <- reactive({
  #   read_excel_file("www/tablePagerank/TCM.xlsx")
  # })
  # 
  # filtered_tcm_data <- reactive({
  #   filter_data(tcm_data(), input$search_input_tcm)
  # })
  # 
  # output$table_tcm <- renderDT({
  #   filtered_tcm_data()
  # })

  # Server logic for T EM tab
  output$tem <- renderText({
    "TF Activity Score: T EM"
  })
  
  # OLD TEM DATA
  # tem_data <- reactive({
  #   read_excel_file("www/tablePagerank/TEM.xlsx")
  # })
  # 
  # filtered_tem_data <- reactive({
  #   filter_data(tem_data(), input$search_input_tem)
  # })
  # 
  # output$table_tem <- renderDT({
  #   filtered_tem_data()
  # })
  # 
  # # Server logic for T RM tab
  # output$trm <- renderText({
  #   "TF Activity Score: T RM"
  # })
  
  # OLD TRM DATA
  # trm_data <- reactive({
  #   read_excel_file("www/tablePagerank/TRM.xlsx")
  # })
  # 
  # filtered_trm_data <- reactive({
  #   filter_data(trm_data(), input$search_input_trm)
  # })
  # 
  # output$table_trm <- renderDT({
  #   filtered_trm_data()
  # })

  # Server logic for Tex Prog tab
  output$texprog <- renderText({
    "TF Activity Score: Tex Prog"
  })
  
  # OLD TEX PROG DATA
  # texprog_data <- reactive({
  #   read_excel_file("www/tablePagerank/TEXprog.xlsx")
  # })
  # 
  # filtered_texprog_data <- reactive({
  #   filter_data(texprog_data(), input$search_input_texprog)
  # })
  # 
  # output$table_texprog <- renderDT({
  #   filtered_texprog_data()
  # })
  

  # Server logic for Tex Eff-like tab
  output$texefflike <- renderText({
    "TF Activity Score: Tex Eff-like"
  })
  
  # OLD TEX EFF LIKE DATA
  # texefflike_data <- reactive({
  #   read_excel_file("www/tablePagerank/TEXeff.xlsx")
  # })
  # 
  # filtered_texefflike_data <- reactive({
  #   filter_data(texefflike_data(), input$search_input_texefflike)
  # })
  # 
  # output$table_texefflike <- renderDT({
  #   filtered_texefflike_data()
  # })

  # Server logic for Tex Term tab
  output$texterm <- renderText({
    "TF Activity Score: Tex Term"
  })
  
  # OLD TEX TERM DATA
  # texterm_data <- reactive({
  #   read_excel_file("www/tablePagerank/TEXterm.xlsx")
  # })
  # 
  # filtered_texterm_data <- reactive({
  #   filter_data(texterm_data(), input$search_input_texterm)
  # })
  # 
  # output$table_texterm <- renderDT({
  #   filtered_texterm_data()
  # })
  
  
  
  #click image code for TF Wave Analysis
  observeEvent(input$c1_link, {
    updateNavbarPage(session, "mainNav", selected = "c1")
  })
  
  observeEvent(input$c2_link, {
    updateNavbarPage(session, "mainNav", selected = "c2")
  })
  
  observeEvent(input$c3_link, {
    updateNavbarPage(session, "mainNav", selected = "c3")
  })
  
  observeEvent(input$c4_link, {
    updateNavbarPage(session, "mainNav", selected = "c4")
  })
  
  observeEvent(input$c5_link, {
    updateNavbarPage(session, "mainNav", selected = "c5")
  })
  
  observeEvent(input$c6_link, {
    updateNavbarPage(session, "mainNav", selected = "c6")
  })
  
  observeEvent(input$c7_link, {
    updateNavbarPage(session, "mainNav", selected = "c7")
  })
  
  
  # Handle specific image part clicks to navigate to the subpages
  observeEvent(input$to_tfcat, {
    updateNavbarPage(session, "mainNav", selected = "tfcatpage")
  })
  
  observeEvent(input$to_tfwave, {
    updateNavbarPage(session, "mainNav", selected = "overview")
  })
  
  observeEvent(input$to_tfnet, {
    updateNavbarPage(session, "mainNav", selected = "tfnetpage")
  })
  
  # SEARCH FOR A TF AND FIND OUT WHICH WAVE THEY ARE APART OF
  # Function to read the file and assign proper column names
  read_searchwave_file <- function(path) {
    df <- read_excel(path)  # Read the Excel file
    colnames(df) <- c("Wave1", "Wave2", "Wave3", "Wave4", "Wave5", "Wave6", "Wave7")  # Customize as needed
    return(df)
    
    return(df)
  }
  
  # Load data reactively
  searchwavedata <- reactive({
    # Provide the path to your Excel file
    read_searchwave_file("www/waveanalysis/searchtfwaves.xlsx")  # Path to your file
  })
  
  # Reactive function to search and filter the data based on the wave search input
  filtered_wave_data <- reactive({
    # Get the search input
    search_term <- input$search_input_wave
    
    # If no search term, return the entire dataset in the desired transposed format
    if (is.null(search_term) || search_term == "") {
      # Initialize an empty list to store the gene names for each wave
      wave_genes <- list()
      
      # Iterate over each wave column and get the associated genes
      for (col in colnames(searchwavedata())) {
        wave_genes[[col]] <- searchwavedata()[[col]]  # Store all gene names for each wave
      }
      
      # Create a data frame with waves as columns and genes as rows
      result_df <- data.frame(wave_genes)
      
      # Return the transposed data frame with wave numbers as column headers and genes as rows
      return(result_df)
    }
    
    # Initialize an empty list to store the result
    result <- list()
    
    # Iterate through each column and check if the search term exists
    for (col in colnames(searchwavedata())) {
      # Filter genes for each column where the search term matches
      matching_genes <- searchwavedata()[[col]][grepl(search_term, searchwavedata()[[col]], ignore.case = TRUE)]
      
      # If there are matching genes, store them
      if (length(matching_genes) > 0) {
        result[[col]] <- paste(matching_genes, collapse = ", ")  # Combine matching genes as a string
      }
    }
    
    # If no results are found, return an empty dataframe
    if (length(result) == 0) {
      return(data.frame(Wave = character(0), Gene = character(0)))  # Return empty data frame if nothing matches
    }
    
    # Convert the result list to a data frame
    result_df <- data.frame(
      Wave = names(result),  # Column names as 'Wave'
      Gene = unlist(result),  # Concatenated list of matching gene names
      stringsAsFactors = FALSE
    )
    
    # Remove any duplicate wave names (keep only one occurrence per wave)
    result_df <- result_df[!duplicated(result_df$Wave), ]
    
    # Remove rows where Gene column is NA or empty
    result_df <- result_df[!is.na(result_df$Gene) & result_df$Gene != "", ]
    
    # Create a matrix with wave names as columns and gene names as rows
    transposed_df <- matrix(unlist(result_df$Gene), nrow = 1)  # Convert the gene names to a row
    
    # Set the column names to the wave numbers, replacing "Wave 1" instead of "wave.1"
    colnames(transposed_df) <- paste("Wave", seq_along(result_df$Wave))
    
    # Return the transposed result
    return(as.data.frame(transposed_df))
  })
  
  # Render the table output based on the filtered and transposed data
  output$table_wave <- renderDT({
    # Get the filtered and transposed data
    df <- filtered_wave_data()
    
    # Render the data table without row names and disable the default search box
    datatable(df, options = list(searching = FALSE), rownames = FALSE)
  })
  
  
  
  
  # # Function to read Excel files
  # read_regulator_file <- function(path) {
  #   df <- read_excel(path)
  #   colnames(df)[1] <- " "  # Adjust column name as needed
  #   return(df)
  # }
  # 
  # # Load data initially
  # tfregulated_data <- reactive({
  #   read_regulator_file("www/networkanalysis/comp_log2FC_RegulatedData_TRMTEXterm.xlsx")
  # })
  # 
  # # Filtered data based on search input
  # filtered_tfregulated_data <- reactive({
  #   req(tfregulated_data())  # Ensure tfregulated_data() is available
  #   if (is.null(input$search_tfregulated_data) || input$search_tfregulated_data == "") {
  #     return(tfregulated_data())
  #   } else {
  #     # Perform filtering based on input$search_tfregulated_data
  #     # Example filtering logic:
  #     # filtered_data <- tfregulated_data() %>%
  #     #   filter(...)  # Add your filtering logic here
  #     # return(filtered_data)
  #     # Replace the above with your actual filtering logic
  #     return(tfregulated_data())  # Placeholder for now
  #   }
  # })
  # 
  # # Render the DataTable
  # output$table_tfregulated_data <- renderDT({
  #   datatable(filtered_tfregulated_data())
  # })

  
  
  # ORIGINALLY SEARCH TRM TEXterm COORELATION
  # Load Excel file when the app starts
  data_tftfimage <- read_excel("www/TFcorintextrm/TF-TFcorTRMTEX.xlsx")
  
  # Reactive function to filter the data based on the search input (case-insensitive)
  filtered_data_tftfimage <- reactive({
    req(input$search)  # Ensure search input is available
    
    # Convert both the column and search input to lowercase for case-insensitive comparison
    data_filtered <- data_tftfimage[tolower(data_tftfimage[["TF Name"]]) == tolower(input$search), ]
    return(data_filtered)
  })
  

  # Render the first column (Gene Names) as clickable links
  output$gene_list_table <- renderUI({
    tagList(
      lapply(data_tftfimage[[1]], function(gene_name) {  # Assuming the first column contains gene names
        tags$div(
          actionLink(
            inputId = paste0("gene_", gene_name),  # Unique input ID for each gene
            label = gene_name
          ),
          style = "margin-bottom: 10px;"  # Add spacing between links
        )
      })
    )
  })

  # Generate dynamic observers for each gene link
  lapply(data_tftfimage[[1]], function(gene_name) {
    observeEvent(input[[paste0("gene_", gene_name)]], {
      # Find the row corresponding to the clicked gene
      selected_gene_data <- data_tftfimage[data_tftfimage[[1]] == gene_name, ]
      img_src <- selected_gene_data[["TF Merged Graph Path"]]  # Replace with the actual image column name

      # Update the image gallery output with the selected gene's image
      output$image_gallery <- renderUI({
        if (!is.null(img_src) && nchar(img_src) > 0) {
          tags$div(
            style = "text-align: center;",
            tags$img(src = img_src, style = "max-width: 100%; height: auto;"),
            tags$p(gene_name)  # Optionally display the gene name below the image
          )
        } else {
          "No image available for the selected gene."
        }
      })
    })
  })

  # Event to handle search functionality (unchanged from your original code)
  observeEvent(input$search_btn, {
    output$result_table <- renderTable({
      if (nrow(filtered_data_tftfimage()) > 0) {
        filtered_data_tftfimage()[, -which(names(filtered_data_tftfimage()) == "TF Merged Graph Path")]  # Show all columns except the image path
      } else {
        NULL  # If no results, display nothing
      }
    })

    output$image_gallery <- renderUI({
      if (nrow(filtered_data_tftfimage()) > 0) {
        image_list <- lapply(1:nrow(filtered_data_tftfimage()), function(i) {
          img_src <- filtered_data_tftfimage()[[ "TF Merged Graph Path" ]][i]
          tags$div(
            style = "text-align: center; margin-bottom: 20px;",
            tags$img(src = img_src, style = "max-width: 100%; height: auto;")
          )
        })
        do.call(tags$div, image_list)
      } else {
        "TF not found. Please search for a valid TF."
      }
    })
  })
  
  
  
  #tf communities
  # Read the first Excel file directly from the specified path
  datatrm <- read_excel("www/tfcommunities/trmcommunities.xlsx")
  
  # Read the second Excel file directly from the specified path
  datatex <- read_excel("www/tfcommunities/texcommunities.xlsx")
  
  # Render the first table without the search button
  output$trmcom <- renderDT({
    datatable(
      datatrm, 
      options = list(
        lengthChange = FALSE, 
        pageLength = 5,
        searching = FALSE  # Disable the search button
      )
    )
  })
  
  # Render the second table without the search button
  output$texcom <- renderDT({
    datatable(
      datatex, 
      options = list(
        lengthChange = FALSE, 
        pageLength = 5,
        searching = FALSE  # Disable the search button
      )
    )
  })
  
  
  # Load your multiomics file
  multiexcel_data <- read_excel("www/multi-omicsdata.xlsx")
  
  # #Render data table + hyperlinks author's name to DOI
  # output$multiomicsdatatable <- renderDT({
  #   # Transform the "author" column to contain hyperlinks using the "DOI" column
  #   multiexcel_data <- multiexcel_data %>%
  #     mutate(
  #       Author = paste0(
  #         "<a href='", 
  #         DOI,  # Column with the full DOI URLs
  #         "' target='_blank'>", 
  #         Author,  # Column with the display text (e.g., author name)
  #         "</a>"
  #       )
  #     ) %>%
  #     select(-DOI)  # Remove the "DOI" column after linking it to the "author" column
  #   
  #   # Dynamically remove empty columns ("18", "19", etc.)
  #   multiexcel_data <- multiexcel_data %>%
  #     select(where(~ !all(is.na(.)) & !all(. == "")))  # Keep only non-empty columns
  #   
  #   # Render the data table with HTML content and no row names
  #   datatable(multiexcel_data, 
  #             options = list(searching = FALSE), 
  #             rownames = FALSE, 
  #             escape = FALSE)  # Allow HTML rendering
  # })
  
  #Render and hyperlink table + edit size so that everything fits into webpage
  output$multiomicsdatatable <- renderDT({
    # Transform the "author" column to contain hyperlinks using the "DOI" column
    multiexcel_data <- multiexcel_data %>%
      mutate(
        Author = paste0(
          "<a href='", 
          DOI,  # Column with the full DOI URLs
          "' target='_blank'>", 
          Author,  # Column with the display text (e.g., author name)
          "</a>"
        )
      ) %>%
      select(-DOI)  # Remove the "DOI" column after linking it to the "author" column
    
    # Dynamically remove empty columns ("18", "19", etc.)
    multiexcel_data <- multiexcel_data %>%
      select(where(~ !all(is.na(.)) & !all(. == "")))  # Keep only non-empty columns
    
    # Render the data table with fit-to-page options
    datatable(
      multiexcel_data, 
      options = list(
        autoWidth = TRUE,  # Adjust column widths automatically
        scrollX = TRUE,    # Enable horizontal scrolling
        pageLength = 10    # Limit rows displayed per page (adjustable)
      ), 
      rownames = FALSE, 
      escape = FALSE  # Allow HTML rendering for links
    )
  })
  
  # --- START: TaijiChat Message Handling ---
  chat_history <- reactiveVal(list()) # Stores list of lists: list(role="user/assistant", content="message")

  observeEvent(input$user_chat_message, {
    req(input$user_chat_message)
    user_message_text <- trimws(input$user_chat_message)
    print(paste("TaijiChat: Received user_chat_message -", user_message_text))

    if (nzchar(user_message_text)) {
      # Check authentication (implement OAuth later in ui.R)
      # For now, system works without auth, but logs as "anonymous"
      current_user <- session$userData$hf_user
      hf_user_id <- if (!is.null(current_user)) current_user$hf_user_id else "anonymous"

      # Check quota before processing
      if (!is.null(supabase_client) && !is.null(current_user)) {
        quota_result <- check_user_quota(supabase_client, hf_user_id)
        if (!quota_result$has_quota) {
          session$sendCustomMessage(type = "agent_response", message = list(
            text = paste("Token quota exceeded. Used:", quota_result$tokens_used, "Remaining: 0")
          ))
          return()
        }
      }

      current_hist <- chat_history()
      updated_hist_user <- append(current_hist, list(list(role = "user", content = user_message_text)))
      chat_history(updated_hist_user)

      agent_instance_val <- rv_agent_instance()

      # Set user context in agent
      if (!is.null(agent_instance_val) && !is.null(current_user)) {
        supabase_user <- session$userData$supabase_user
        agent_instance_val$set_user_context(
          user_id = supabase_user$id,
          hf_user_id = hf_user_id
        )
      } 

      if (!is.null(agent_instance_val)) {
        # Ensure history is a list of R named lists, then r_to_py will convert to list of Python dicts
        py_hist_for_agent <- lapply(updated_hist_user, function(turn) {
          list(role = turn$role, content = turn$content)
        })
        # py_hist_for_agent_converted <- reticulate::r_to_py(py_hist_for_agent)

        # Send a "Thinking..." message to UI before long computation
        session$sendCustomMessage(type = "agent_thinking_started", message = list(text = "Thinking..."))

        tryCatch({
          print(paste("TaijiChat: Sending to Python agent - Query:", user_message_text))
          # For debugging, convert history to JSON string to see its structure if needed
          # hist_json_debug <- jsonlite::toJSON(py_hist_for_agent, auto_unbox = TRUE)
          # print(paste("TaijiChat: Conversation history (JSON for debug):", hist_json_debug))
          
          # Get literature search preference (default to FALSE if not set)
          literature_enabled <- if (is.null(input$literature_search_enabled)) FALSE else input$literature_search_enabled
          print(paste("TaijiChat: Literature search enabled:", literature_enabled))
          
          # Call Python agent method with literature preference
          # The process_single_query method in Python expects history as a list of dicts.
          # reticulate::r_to_py should handle the conversion of the list of R named lists.
          agent_reply_py <- agent_instance_val$process_single_query_with_preferences(
            user_query_text = user_message_text,
            conversation_history_from_r = py_hist_for_agent, # Pass the R list of lists
            literature_enabled = literature_enabled
          )
          # Explicitly convert potential Python object to R character string
          agent_reply_text <- as.character(agent_reply_py)

          print(paste("TaijiChat: Received from Python agent -", agent_reply_text))
          
          
          # Check if this is an image response
          if (startsWith(agent_reply_text, "TAIJICHAT_IMAGE_RESPONSE:")) {
            # Extract the JSON part - START AFTER THE COLON IN THE PREFIX
            json_str <- substr(agent_reply_text, 26, nchar(agent_reply_text))
            
            # Debug the JSON string
            print(paste("Attempting to parse JSON:", json_str))
            
            # Try parsing in a safer way
            tryCatch({
              # Remove any leading/trailing whitespace and ensure proper JSON format
              json_str <- trimws(json_str)
              
              # Try to parse the JSON
              image_info <- NULL
              if (nchar(json_str) > 0) {
                # Attempt to parse using various approaches
                image_info <- tryCatch({
                  jsonlite::fromJSON(json_str)
                }, error = function(e1) {
                  tryCatch({
                    # Try unescaping first
                    jsonlite::fromJSON(gsub('\\\\"', '"', json_str))
                  }, error = function(e2) {
                    NULL
                  })
                })
              }
              
              if (!is.null(image_info) && !is.null(image_info$image_path)) {
                # Debug image path
                image_path_original <- image_info$image_path
                image_path_normalized <- sub("^www/", "", image_info$image_path)
                
                # Check if file exists
                file_exists_check <- file.exists(image_path_original)
                print(paste("Image path debug - Original:", image_path_original, 
                           "Normalized:", image_path_normalized,
                           "File exists:", file_exists_check))
                
                # Add a special marker to the message to trigger image display in UI
                image_html <- paste0(
                  '<div class="chat-image-container">',
                  '<img src="', image_path_normalized, '" class="chat-image-preview" onclick="showFullImage(\'', 
                  image_path_normalized, '\')">',
                  '</div>'
                )
                
                # Get original response or use a default
                original_response <- ifelse(!is.null(image_info$original_response), 
                                          image_info$original_response, 
                                          "I've analyzed this image.")
                
                # Combine the image HTML with the original text response
                enhanced_reply <- paste0(
                  image_html,
                  "<br>",
                  original_response
                )
                
                agent_reply_text <- enhanced_reply
                
                # Send a custom message to ensure the image display script is active
                session$sendCustomMessage(type = "activate_image_viewer", message = list())
              } else {
                warning("Failed to extract image path from JSON")
                agent_reply_text <- paste("I analyzed an image but had trouble displaying it. Here's what I found:", 
                                       gsub("TAIJICHAT_IMAGE_RESPONSE:.*", "", agent_reply_text))
              }
            }, error = function(e) {
              warning(paste("JSON parsing error:", e$message, "- JSON string:", json_str))
              print(paste("JSON parsing error:", e$message, "- JSON string:", json_str))
              # Just use the original text in case of parsing error
              agent_reply_text <- paste("I analyzed an image but had trouble displaying it. Here's what I found:", 
                                       substr(agent_reply_text, 25, nchar(agent_reply_text)))
            })
          }

          final_hist <- append(updated_hist_user, list(list(role = "assistant", content = agent_reply_text)))
          chat_history(final_hist)

          session$sendCustomMessage(type = "agent_chat_response", message = list(text = agent_reply_text))

          # Update quota display after successful query
          if (!is.null(supabase_client) && !is.null(current_user)) {
            quota_result <- check_user_quota(supabase_client, hf_user_id)
            tokens_used <- quota_result$tokens_used
            token_quota <- session$userData$supabase_user$token_quota %||% 100000
            quota_text <- paste(tokens_used, "/", token_quota)
            session$sendCustomMessage('update_quota', list(quota = quota_text))
          }

        }, error = function(e) {
          error_message <- paste("TaijiChat: Error calling Python agent or processing response:", e$message)
          warning(error_message)
          print(error_message)
          session$sendCustomMessage(type = "agent_chat_response", message = list(text = paste("Sorry, an error occurred with the agent.")))
        })
      } else {
        warning("TaijiChat: Agent instance is NULL. Cannot process chat message.")
        print("TaijiChat: Agent instance is NULL. Cannot process chat message.")
        session$sendCustomMessage(type = "agent_chat_response", message = list(text = "The chat agent is not available. Please check server logs."))
      }
    } else {
      print("TaijiChat: Received empty user_chat_message.")
    }
  })
  
  
  # --- END: TaijiChat Message Handling ---

  #Render and hyperlink table + edit size so that everything fits into webpage
  output$multiomicsdatatable <- renderDT({
    # Transform the "author" column to contain hyperlinks using the "DOI" column
    multiexcel_data <- multiexcel_data %>%
      mutate(
        Author = paste0(
          "<a href='", 
          DOI,  # Column with the full DOI URLs
          "' target='_blank'>", 
          Author,  # Column with the display text (e.g., author name)
          "</a>"
        )
      ) %>%
      select(-DOI)  # Remove the "DOI" column after linking it to the "author" column
    
    # Dynamically remove empty columns ("18", "19", etc.)
    multiexcel_data <- multiexcel_data %>%
      select(where(~ !all(is.na(.)) & !all(. == "")))  # Keep only non-empty columns
    
    # Render the data table with fit-to-page options
    datatable(
      multiexcel_data, 
      options = list(
        autoWidth = TRUE,  # Adjust column widths automatically
        scrollX = TRUE,    # Enable horizontal scrolling
        pageLength = 10    # Limit rows displayed per page (adjustable)
      ), 
      rownames = FALSE, 
      escape = FALSE  # Allow HTML rendering for links
    )
  })
  
  # Update reactive expressions to use warning overlay
  output$tfData <- renderDT({
    withWarningOverlayReactive(session, {
      # Your existing reactive code here
      # This will automatically show the warning overlay for long-running operations
    }, "get_processed_tf_data")
  })
}


# # ############################## CODE GRAVEYARD ##############################