ui.R

# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# ui.R - User Interface
#
# Description:
# This file defines the user interface for the Shiny application. It lays out
# all the tabs, sidebars, inputs, and output placeholders that the user will
# interact with. It sources objects from global.R implicitly.
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

ui <- navbarPage(
  title = div(
    img(
      src = "app_icon.png",
      height = "30px",
      style = "margin-top: -5px; margin-right: 10px;"
    ),
    "JABS Data Modeler"
  ),
  theme = shinytheme("yeti"),

  # --- TAB 1: Data Exploration & Preprocessing ---
  tabPanel(
    "1. Data Exploration & Preprocessing",
    sidebarLayout(
      sidebarPanel(
        width = 3,
        div(
          style = "text-align: center;",
          img(src = "app_icon.png", height = "150px"),
          hr()
        ),
        h4("1. Load Data"),
        fileInput(
          "data_file",
          "Upload CSV Data File",
          accept = c(
            "text/csv",
            "text/comma-separated-values,text/plain",
            ".csv"
          )
        ),
        uiOutput("start_col_selector_ui"),
        uiOutput("end_col_selector_ui"),
        tags$hr(),
        
         h4("2. Select Data Subsets"),
         checkboxGroupInput(
           "tx_selector",
           "Select Treatment Doses (Tx):",
           choices = NULL,
           selected = NULL
         ),
         div(
           style = "margin-bottom: 10px;",
           actionButton("select_all_timepoints", "Select All", class = "btn-sm btn-outline-primary", style = "margin-right: 5px;"),
           actionButton("select_none_timepoints", "Select None", class = "btn-sm btn-outline-secondary")
         ),
         checkboxGroupInput(
           "timepoint_selector",
           "Select Timepoints:",
           choices = NULL,
           selected = NULL
         ),
        tags$hr(),
        
        h4("3. Preprocessing Options"),
         checkboxInput(
           "remove_zero_var",
           "Remove features with zero variance",
           value = TRUE
         ),
        selectInput(
          "outlier_method",
          "Outlier Detection Method:",
          choices = list(
            "None" = "none",
            "IQR Method" = "iqr",
            "Z-Score Method" = "zscore"
          ),
          selected = "none"
        ),
        selectInput(
          "outlier_action",
          "Outlier Action:",
          choices = list(
            "Flag Only" = "flag",
            "Remove Outliers" = "remove"
          ),
          selected = "flag"
        ),
        selectInput(
          "transform_method",
          "Data Transformation:",
          choices = list(
            "None" = "none",
            "Log Transform" = "log",
            "Square Root" = "sqrt",
            "Box-Cox" = "boxcox"
          ),
          selected = "none"
        ),
        actionButton("run_preprocessing", "Run Preprocessing", class = "btn-primary"),
        tags$hr(),

        # Plotting controls removed - using direct controls in main panel
      ),
      mainPanel(
        width = 9,
        
        # Add comprehensive workflow instructions
        div(
          style = "background-color: #e8f4fd; padding: 20px; border-radius: 5px; margin-bottom: 20px;",
          h3("📋 Data Analysis Workflow Instructions", style = "color: #2c3e50; margin-top: 0;"),
          
          h4("Step 1: Load and Select Data", style = "color: #34495e; font-weight: bold;"),
          tags$ol(
            tags$li("Upload your CSV data file using the file input on the left"),
            tags$li("Select the first and last feature columns to define your analysis range"),
            tags$li("Choose which treatment doses (Tx) and timepoints to include in your analysis")
          ),
          
          h4("Step 2: Preprocessing Decisions", style = "color: #34495e; font-weight: bold;"),
          tags$ol(
            tags$li(tags$strong("Remove Zero Variance Features:"), " Check this to remove features with no variation across samples"),
            tags$li(tags$strong("Outlier Detection:"), " Choose between 'None', 'IQR', or 'Z-score' methods to identify outliers"),
            tags$li(tags$strong("Outlier Removal:"), " Decide whether to remove detected outliers or just flag them"),
            tags$li(tags$strong("Data Transformation:"), " Select 'None', 'Log', 'Square Root', or 'Box-Cox' if your data needs transformation")
          ),
          
          h4("Step 3: Run Preprocessing", style = "color: #34495e; font-weight: bold;"),
          tags$ol(
            tags$li("Click 'Run Preprocessing' to apply your selected preprocessing steps"),
            tags$li("Review the preprocessing log to see what changes were made"),
            tags$li("Examine the feature visualization plots to assess data quality")
          ),
          
          h4("Step 4: Continue to Analysis Tabs", style = "color: #34495e; font-weight: bold;"),
          tags$ol(
            tags$li(tags$strong("Tab 2 - Heatmaps:"), " Explore feature correlations and relationships"),
            tags$li(tags$strong("Tab 3 - Data Tables:"), " View detailed data summaries and statistics"),
            tags$li(tags$strong("Tab 4 - PCA:"), " Perform Principal Component Analysis to reduce dimensionality"),
            tags$li(tags$strong("Tab 5 - LDA:"), " Use Linear Discriminant Analysis to classify treatment groups")
          ),
          
          div(
            style = "background-color: #fff3cd; padding: 10px; border-radius: 3px; margin-top: 15px;",
            tags$strong("💡 Tip:", style = "color: #856404;"),
            " Complete preprocessing in this tab before moving to PCA and LDA analysis tabs. The analysis tabs will automatically use your preprocessed data."
          )
        ),
        
        h3("Preprocessing Log & Summary"),
        verbatimTextOutput("preprocessing_summary"),
      )
    )
  ),

  # --- TAB 2: Correlation Analysis (Heatmaps) ---
  tabPanel(
    "2. Correlation Analysis",
    sidebarLayout(
      sidebarPanel(
        width = 3,
        h4("Correlation Controls"),
        conditionalPanel(
          condition = "!output.preprocessed_data_exists",
          div(
            class = "alert alert-warning",
            "Please go to Tab 1 and click 'Run Preprocessing' to process your data before using this tab."
          )
        ),
        conditionalPanel(
          condition = "output.preprocessed_data_exists",
          p("Use the controls below to select features for correlation analysis."),
          textInput("corr_feature_search", "Search features:"),
          actionButton("select_all_corr_features", "Select All Filtered", class = "btn-sm btn-outline-primary", style = "margin-right: 5px;"),
          actionButton("clear_all_corr_features", "Clear Selection", class = "btn-sm btn-outline-secondary"),
          tags$label("Features:", `for` = "correlation_features_checkboxes"),
          uiOutput("correlation_feature_selector_ui"),
          tags$hr()
        )
      ),
      mainPanel(
        width = 9,
        h3("Feature Analysis"),
        tabsetPanel(
          tabPanel(
            "Heatmap",
            p(
              "The heatmap shows the Pearson correlation between all selected features. Features are automatically clustered by similarity."
            ),
            actionButton("run_correlation_btn", "Generate Correlation Analysis", class = "btn-primary", style = "margin-bottom: 15px;"),
            fluidRow(
              column(
                width = 4,
                selectInput(
                  "correlation_heatmap_colors",
                  "Color Scale:",
                  choices = list(
                    "Blue-White-Red" = "RdBu",
                    "Red-Yellow-Blue" = "RdYlBu", 
                    "Purple-White-Green" = "PRGn",
                    "Blue-White-Yellow" = "RdYlBu",
                    "Red-White-Blue" = "RdBu_r",
                    "Viridis" = "viridis",
                    "Plasma" = "plasma",
                    "Inferno" = "inferno",
                    "Magma" = "magma",
                    "Spectral" = "Spectral",
                    "Coolwarm" = "coolwarm",
                    "RdBu (Reversed)" = "RdBu_r",
                    "RdYlBu (Reversed)" = "RdYlBu_r"
                  ),
                  selected = "RdBu"
                )
              ),
              column(
                width = 4,
                numericInput(
                  "correlation_heatmap_font_size",
                  "Font Size:",
                  value = 12,
                  min = 6,
                  max = 24,
                  step = 1
                )
              )
            ),
            plotOutput("correlation_heatmap_output", height = "800px"),
            tags$hr(),
            h5("Download Options"),
            fluidRow(
              column(
                width = 3,
                numericInput(
                  "correlation_pdf_width",
                  "PDF Width (inches):",
                  value = 12,
                  min = 6,
                  max = 20,
                  step = 1
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_pdf_height",
                  "PDF Height (inches):",
                  value = 10,
                  min = 6,
                  max = 20,
                  step = 1
                )
              ),
              column(
                width = 3,
                downloadButton("download_heatmap_pdf", "Download Heatmap PDF", class = "btn-success")
              )
            )
          ),
          tabPanel(
            "Two-Feature Scatter Plot",
            p(
              "Select two features to visualize their relationship. Customize the plot with interactive controls for faceting, color, and shape."
            ),
            fluidRow(
              column(
                width = 3,
                selectInput(
                  "correlation_feature_x",
                  "X-axis Feature:",
                  choices = NULL
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_feature_y",
                  "Y-axis Feature:",
                  choices = NULL
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_facet_by",
                  "Facet By:",
                  choices = list(
                    "None" = "none",
                    "Timepoint" = "Timepoint",
                    "Treatment" = "Tx",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "Timepoint"
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_color_by",
                  "Color By:",
                  choices = list(
                    "Treatment" = "Tx",
                    "Timepoint" = "Timepoint",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "Tx"
                )
              )
            ),
            fluidRow(
              column(
                width = 3,
                selectInput(
                  "correlation_shape_by",
                  "Shape By:",
                  choices = list(
                    "None" = "none",
                    "Treatment" = "Tx",
                    "Timepoint" = "Timepoint",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "none"
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_point_size",
                  "Point Size:",
                  value = 2,
                  min = 0.5,
                  max = 10,
                  step = 0.5
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_fit_line",
                  "Fit Line:",
                  choices = list(
                    "None" = "none",
                    "Linear" = "lm",
                    "Smooth" = "loess"
                  ),
                  selected = "none"
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_line_type",
                  "Line Type:",
                  choices = list(
                    "Solid" = "solid",
                    "Dashed" = "dashed",
                    "Dotted" = "dotted",
                    "Dot-dash" = "dotdash",
                    "Long dash" = "longdash",
                    "Two dash" = "twodash"
                  ),
                  selected = "solid"
                )
              )
            ),
            fluidRow(
              column(
                width = 3,
                numericInput(
                  "correlation_line_width",
                  "Line Width:",
                  value = 1,
                  min = 0.1,
                  max = 5,
                  step = 0.1
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_alpha",
                  "Transparency:",
                  value = 0.7,
                  min = 0.1,
                  max = 1,
                  step = 0.1
                )
              ),
              column(
                width = 3,
                checkboxInput(
                  "correlation_show_legend",
                  "Show Legend",
                  value = TRUE
                )
              ),
              column(
                width = 3,
                checkboxInput(
                  "correlation_show_grid",
                  "Show Grid",
                  value = TRUE
                )
              )
            ),
            fluidRow(
              column(
                width = 3,
                selectInput(
                  "correlation_theme",
                  "Plot Theme:",
                  choices = list(
                    "Minimal" = "minimal",
                    "Classic" = "classic",
                    "Gray" = "gray",
                    "Light" = "light",
                    "Dark" = "dark",
                    "B&W" = "bw",
                    "Void" = "void"
                  ),
                  selected = "minimal"
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_2d_font_size",
                  "Font Size:",
                  value = 14,
                  min = 8,
                  max = 24,
                  step = 1
                )
              )
            ),
            plotOutput("correlation_scatter_output", height = "600px"),
            tags$hr(),
            h5("Download Options"),
            fluidRow(
              column(
                width = 3,
                numericInput(
                  "correlation_2d_pdf_width",
                  "PDF Width (inches):",
                  value = 10,
                  min = 6,
                  max = 20,
                  step = 1
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_2d_pdf_height",
                  "PDF Height (inches):",
                  value = 8,
                  min = 6,
                  max = 20,
                  step = 1
                )
              ),
              column(
                width = 3,
                downloadButton("download_2d_pdf", "Download 2D Plot PDF", class = "btn-success")
              )
            )
          ),
          tabPanel(
            "1D Scatter + Box Plot",
            p(
              "Visualize the distribution of a single feature using both scatter points and box plots. Use the same faceting and coloring options as the 2D scatter plot."
            ),
            fluidRow(
              column(
                width = 3,
                selectInput(
                  "correlation_1d_x_axis",
                  "X-axis (Grouping):",
                  choices = list(
                    "Treatment" = "Tx",
                    "Timepoint" = "Timepoint",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "Timepoint"
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_1d_feature",
                  "Feature to Plot:",
                  choices = NULL
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_1d_facet_by",
                  "Facet By:",
                  choices = list(
                    "None" = "none",
                    "Timepoint" = "Timepoint",
                    "Treatment" = "Tx",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "none"
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_1d_color_by",
                  "Color By:",
                  choices = list(
                    "Treatment" = "Tx",
                    "Timepoint" = "Timepoint",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "Tx"
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_1d_shape_by",
                  "Shape By:",
                  choices = list(
                    "None" = "none",
                    "Treatment" = "Tx",
                    "Timepoint" = "Timepoint",
                    "Sex" = "Sex",
                    "Cohort" = "Cohort",
                    "Experiment Number" = "ExptNumber",
                    "Pen ID" = "Pen_ID",
                    "LL" = "LL"
                  ),
                  selected = "none"
                )
              )
            ),
            fluidRow(
              column(
                width = 3,
                numericInput(
                  "correlation_1d_point_size",
                  "Point Size:",
                  value = 2,
                  min = 0.5,
                  max = 10,
                  step = 0.5
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_1d_alpha",
                  "Transparency:",
                  value = 0.7,
                  min = 0.1,
                  max = 1,
                  step = 0.1
                )
              ),
              column(
                width = 3,
                checkboxInput(
                  "correlation_1d_show_legend",
                  "Show Legend",
                  value = TRUE
                )
              ),
              column(
                width = 3,
                checkboxInput(
                  "correlation_1d_show_boxplot",
                  "Show Box Plot",
                  value = TRUE
                )
              )
            ),
            fluidRow(
              column(
                width = 3,
                sliderInput(
                  "correlation_1d_jitter_width",
                  "Point Jitter:",
                  min = 0.0,
                  max = 1.0,
                  value = 0.4,
                  step = 0.1
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_1d_font_size",
                  "Font Size:",
                  value = 14,
                  min = 8,
                  max = 24,
                  step = 1
                )
              ),
              column(
                width = 3,
                selectInput(
                  "correlation_1d_theme",
                  "Plot Theme:",
                  choices = list(
                    "Minimal" = "minimal",
                    "Classic" = "classic",
                    "Gray" = "gray",
                    "Light" = "light",
                    "Dark" = "dark",
                    "B&W" = "bw",
                    "Void" = "void"
                  ),
                  selected = "minimal"
                )
              )
            ),
            plotOutput("correlation_1d_output", height = "600px"),
            tags$hr(),
            h5("Download Options"),
            fluidRow(
              column(
                width = 3,
                numericInput(
                  "correlation_1d_pdf_width",
                  "PDF Width (inches):",
                  value = 10,
                  min = 6,
                  max = 20,
                  step = 1
                )
              ),
              column(
                width = 3,
                numericInput(
                  "correlation_1d_pdf_height",
                  "PDF Height (inches):",
                  value = 8,
                  min = 6,
                  max = 20,
                  step = 1
                )
              ),
              column(
                width = 3,
                downloadButton("download_1d_pdf", "Download 1D Plot PDF", class = "btn-success")
              )
            )
          )
        )
      )
    )
  ),

  # --- TAB 3: Data Summary Tables ---
  tabPanel(
    "3. Data Summary",
    sidebarLayout(
      sidebarPanel(
        width = 3,
        h4("Summary Controls"),
        conditionalPanel(
          condition = "!output.preprocessed_data_exists",
          div(
            class = "alert alert-warning",
            "Please go to Tab 1 and click 'Run Preprocessing' to process your data before using this tab."
          )
        ),
        conditionalPanel(
          condition = "output.preprocessed_data_exists",
          p("Generate summary tables based on metadata columns from your data."),
          tags$hr(),
          h5("Row Variables"),
          p("Select variables to group by (rows in the summary table):"),
        uiOutput("summary_row_selector_ui"),
          tags$hr(),
          h5("Column Variables (Optional)"),
          p("Select variables to create columns (optional - leave empty for simple counts):"),
        uiOutput("summary_col_selector_ui"),
          tags$hr(),
          h5("Export Table"),
        p("Download the generated summary table in your desired format."),
        downloadButton(
          "download_summary_csv",
          "Download as CSV",
          class = "btn-info"
        ),
        br(),
        br(),
        downloadButton(
          "download_summary_latex",
          "Download as LaTeX",
          class = "btn-info"
          )
        )
      ),
      mainPanel(
        width = 9,
        h3("Data Summary Table"),
        p(
          "This table shows summary statistics based on the selected row and column variables. The table will update automatically when you change the variable selections."
        ),
        p(
          tags$strong("Margins:"), "Row totals show the sum across all columns for each row. Column totals show the sum across all rows for each column. The bottom-right cell shows the grand total."
        ),
        dataTableOutput("summary_table_output")
      )
    )
  ),

  # --- TAB 4: Principal Component Analysis (PCA) ---
  tabPanel(
    "4. Principal Component Analysis (PCA)",
    sidebarLayout(
      sidebarPanel(
        width = 3,
        h4("PCA Controls"),
        conditionalPanel(
          condition = "!output.preprocessed_data_exists",
          div(
            class = "alert alert-warning",
            "Please go to Tab 1 and click 'Run Preprocessing' to process your data before using this tab."
          )
        ),
        conditionalPanel(
          condition = "output.preprocessed_data_exists",
          p("PCA will be performed on the preprocessed features from Tab 1 for each timepoint."),
          tags$hr(),
          h5("Feature Selection Method"),
          radioButtons(
            "pca_feature_method",
            "Choose how to select features:",
            choices = list(
              "By Variance Explained (%)" = "variance",
              "By Number of Components" = "components"
            ),
            selected = "variance"
          ),
          conditionalPanel(
            condition = "input.pca_feature_method == 'variance'",
            numericInput(
              "pca_variance_threshold",
              "Variance Explained (%):",
              value = 95,
              min = 50,
              max = 99.9,
              step = 0.1
            )
          ),
          conditionalPanel(
            condition = "input.pca_feature_method == 'components'",
            sliderInput(
              "pca_num_components",
              "Number of Components:",
              min = 1,
              max = 50,
              value = 10,
              step = 1
            )
          ),
          actionButton("run_pca_btn", "Run PCA Analysis", class = "btn-primary"),
          tags$hr(),
          h5("Plot Controls"),
          selectInput(
            "pca_plot_type",
            "Select Plot Type:",
            choices = list(
              "Scree Plot" = "scree",
              "PC Scatter Plot" = "scatter",
              "PC Loadings Plot" = "loadings"
            ),
            selected = "scree"
          ),
          conditionalPanel(
            condition = "input.pca_plot_type == 'scatter'",
            selectInput(
              "pca_pc_x",
              "X-axis PC:",
              choices = list("PC1" = "PC1", "PC2" = "PC2"),
              selected = "PC1"
            ),
            selectInput(
              "pca_pc_y",
              "Y-axis PC:",
              choices = list("PC1" = "PC1", "PC2" = "PC2"),
              selected = "PC2"
            )
          ),
          conditionalPanel(
            condition = "input.pca_plot_type == 'loadings'",
            selectInput(
              "pca_loadings_pc",
              "PC for Loadings:",
              choices = list("PC1" = "PC1", "PC2" = "PC2"),
              selected = "PC1"
            ),
            numericInput(
              "pca_loadings_n_features",
              "Number of Top Features:",
              value = 20,
              min = 5,
              max = 50,
              step = 5
            )
          ),
          tags$hr(),
          h5("Download Options"),
          numericInput(
            "pca_pdf_width",
            "PDF Width (inches):",
            value = 12,
            min = 6,
            max = 20,
            step = 1
          ),
          numericInput(
            "pca_pdf_height",
            "PDF Height (inches):",
            value = 10,
            min = 6,
            max = 20,
            step = 1
          ),
          downloadButton("download_pca_pdf", "Download PDF", class = "btn-success")
        )
      ),
      mainPanel(
        width = 9,
        h3("PCA Analysis Results"),
        conditionalPanel(
          condition = "output.preprocessed_data_exists",
          p("Principal Component Analysis performed on preprocessed features for each timepoint."),
          conditionalPanel(
            condition = "output.pca_results_available",
            h4("PCA Results Summary"),
            uiOutput("pca_results_ui"),
            tags$hr(),
            plotOutput("pca_main_plot", height = "600px")
          ),
          conditionalPanel(
            condition = "!output.pca_results_available",
            div(
              style = "text-align: center; padding: 50px;",
              h4("No PCA results available"),
              p("Please run PCA analysis first using the controls in the sidebar.")
            )
          )
        )
      )
    )
  ),

  # --- TAB 5: Linear Discriminant Analysis (LDA) ---
  tabPanel(
    "5. Linear Discriminant Analysis (LDA)",
    sidebarLayout(
      sidebarPanel(
        width = 3,
        h4("LDA Controls"),
        p("LDA will automatically use the PCs that account for 95% of the variance (as selected in the PCA tab) to discriminate between treatment doses (Tx) for each timepoint."),
        
        tags$hr(),
        h5("LDA Parameters"),
        numericInput(
          "lda_cv_folds",
          "Cross-Validation Folds:",
          value = 5,
          min = 3,
          max = 10,
          step = 1
        ),
        actionButton("run_lda_btn", "Run LDA Analysis", class = "btn-primary"),
        tags$hr(),
        h5("Plot Controls"),
        selectInput(
          "lda_plot_type",
          "Select Plot Type:",
          choices = list(
            "LDA Scatter Plot" = "scatter",
            "LDA Univariate Plot" = "univariate",
            "Accuracy Comparison" = "accuracy"
          ),
          selected = "scatter"
        ),
        conditionalPanel(
          condition = "input.lda_plot_type == 'scatter'",
          div(
            style = "background-color: #fff3cd; padding: 10px; border-radius: 3px; margin-bottom: 10px;",
            tags$small(
              "Note: With 2 dose groups, only LD1 exists. The plot will automatically show a 1D visualization.",
              style = "color: #856404;"
            )
          ),
          selectInput(
            "lda_ld_x",
            "X-axis LD:",
            choices = list("LD1" = "LD1", "LD2" = "LD2"),
            selected = "LD1"
          ),
          selectInput(
            "lda_ld_y",
            "Y-axis LD:",
            choices = list("LD1" = "LD1", "LD2" = "LD2"),
            selected = "LD2"
          ),
          checkboxInput(
            "lda_show_confidence",
            "Show Confidence Intervals",
            value = TRUE
          )
        ),
        conditionalPanel(
          condition = "input.lda_plot_type == 'univariate'",
          selectInput(
            "lda_ld_univariate",
            "LD Axis:",
            choices = list("LD1" = "LD1", "LD2" = "LD2"),
            selected = "LD1"
          ),
          checkboxInput(
            "lda_show_density",
            "Show Density Plot (vs Box Plot)",
            value = TRUE
          )
        ),
        tags$hr(),
        h5("Download Options"),
        numericInput(
          "lda_pdf_width",
          "PDF Width (inches):",
          value = 12,
          min = 6,
          max = 20,
          step = 1
        ),
        numericInput(
          "lda_pdf_height",
          "PDF Height (inches):",
          value = 10,
          min = 6,
          max = 20,
          step = 1
        ),
        downloadButton("download_lda_pdf", "Download PDF", class = "btn-success")
      ),
      mainPanel(
        width = 9,
        h3("LDA Analysis Results"),
        
        # Add comprehensive explanations below controls
        div(
          style = "background-color: #f8f9fa; padding: 15px; border-radius: 5px; margin-bottom: 20px;",
          h4("Understanding LDA Results", style = "color: #2c3e50; margin-top: 0;"),
          
          h5("Accuracy Metrics:", style = "color: #34495e; font-weight: bold;"),
          tags$ul(
            tags$li(tags$strong("Overall Accuracy:"), " Percentage of samples correctly classified using all available data. This is the 'best case' scenario."),
            tags$li(tags$strong("CV Accuracy:"), " Cross-validation accuracy - average accuracy when the model is tested on data it hasn't seen during training. This is more realistic."),
            tags$li(tags$strong("CV Standard Deviation:"), " How much the CV accuracy varies across different test sets. Lower values indicate more consistent performance.")
          ),
          
          h5("Expected Values by Chance:", style = "color: #34495e; font-weight: bold;"),
          tags$ul(
            tags$li(tags$strong("Random Guessing:"), " With N dose groups, random guessing would achieve ~100/N% accuracy"),
            tags$li(tags$strong("Examples:"), " 2 groups = 50% chance, 3 groups = 33% chance, 4 groups = 25% chance"),
            tags$li(tags$strong("Note:"), " The actual number of dose groups depends on your data selection in Tab 1"),
            tags$li(tags$strong("Good Performance:"), " >50% accuracy suggests the model is learning meaningful patterns"),
            tags$li(tags$strong("Excellent Performance:"), " >75% accuracy indicates strong dose discrimination"),
            tags$li(tags$strong("Perfect Performance:"), " 100% accuracy means perfect separation (rare in real data)")
          ),
          
          h5("Interpreting Results:", style = "color: #34495e; font-weight: bold;"),
          tags$ul(
            tags$li("If CV accuracy is much lower than overall accuracy, the model may be overfitting"),
            tags$li("High CV standard deviation suggests the model's performance is inconsistent"),
            tags$li("LDA plots show how well the doses are separated in the reduced dimensional space"),
            tags$li("Confidence intervals show the uncertainty in dose group boundaries")
          ),
          
          h5("Plot Types:", style = "color: #34495e; font-weight: bold;"),
          tags$ul(
            tags$li(tags$strong("2D Scatter Plot:"), " Shows dose separation when 3+ dose groups are selected"),
            tags$li(tags$strong("1D Scatter Plot:"), " Shows LD1 values as jittered points when only 2 dose groups are selected"),
            tags$li(tags$strong("Note:"), " With 2 dose groups, only 1 discriminant function (LD1) exists mathematically")
          )
        ),
        
        conditionalPanel(
          condition = "output.lda_results_available",
          h4("Analysis Summary"),
          div(
            style = "background-color: #e8f4fd; padding: 15px; border-radius: 5px; margin-bottom: 20px;",
            uiOutput("lda_interpretation")
          ),
          h4("Detailed Results by Timepoint"),
          uiOutput("lda_results_ui")
        ),
        conditionalPanel(
          condition = "!output.lda_results_available",
          div(
            style = "text-align: center; padding: 50px;",
            h4("No LDA results available"),
            p("Please run LDA analysis first using the controls in the sidebar.")
          )
        )
      )
    )
  )
)