Module Server for Sample Variable Specification

This defines the server part for the sample variable specification.

Usage

sampleVarSpecServer(
  id,
  experiment_name,
  original_data,
  transformed_data = original_data,
  assign_lists = reactiveValues(),
  num_levels = NULL,
  categorical_only = !is.null(num_levels),
  explicit_na = FALSE,
  label_modal_title = "Please click to group the original factor levels"
)

Arguments

id

(string) the shiny module id.

experiment_name

(reactive string)
name of the input experiment.

original_data

(reactive SummarizedExperiment)
input experiment where the sample variables extracted via SummarizedExperiment::colData() should be eligible for selection.

transformed_data

(reactive SummarizedExperiment)
used when multiple sample variables can be selected in the app. In that case, pass here the pre-transformed data.

assign_lists

(reactivevalues)
object to share factor level groupings across multiple sample variables.

num_levels

(count or NULL)
required number of levels after combining original levels. If NULL then all numbers of levels are allowed.

categorical_only

(flag)
whether only categorical variables should be selected from.

explicit_na

(flag)
whether the colData of original_data will be transformed with hermes::h_df_factors_with_explicit_na before further processing. That means also that NA will be made an explicit factor level and counted for num_levels.

label_modal_title

(string)
title for the dialog that asks for the text input.

Value

Reactive SummarizedExperiment::SummarizedExperiment which can be used as input for the relevant hermes functions.

Note

Only atomic columns (e.g. not DataFrame columns) of the colData which are not completely missing (NA) will be shown for selection. If num_levels is specified then only factor columns will be available.

See also

sampleVarSpecInput() for the module UI.

Examples

ui <- function(id,
               datasets) {
  ns <- NS(id)
  mae <- datasets$get_data("MAE", filtered = FALSE)
  experiment_name_choices <- names(mae)
  teal.widgets::standard_layout(
    encoding = div(
      selectInput(ns("experiment_name"), "Select experiment", experiment_name_choices),
      sampleVarSpecInput(ns("facet_var"), "Select faceting variable")
    ),
    output = plotOutput(ns("plot"))
  )
}
server <- function(id,
                   datasets) {
  moduleServer(id, function(input, output, session) {
    experiment_data <- reactive({
      req(input$experiment_name)
      mae <- datasets$get_data("MAE", filtered = TRUE)
      object <- mae[[input$experiment_name]]
      SummarizedExperiment::colData(object) <- hermes::df_cols_to_factor(SummarizedExperiment::colData(object))
      object
    })
    facet_var_spec <- sampleVarSpecServer(
      "facet_var",
      experiment_name = reactive({
        input$experiment_name
      }),
      original_data = experiment_data
    )
    output$plot <- renderPlot({
      experiment_data_final <- facet_var_spec$experiment_data()
      facet_var <- facet_var_spec$sample_var()
      hermes::draw_boxplot(
        experiment_data_final,
        assay_name = "counts",
        genes = hermes::gene_spec(hermes::genes(experiment_data_final)[1]),
        facet_var = facet_var
      )
    })
  })
}
my_app <- function() {
  mae <- hermes::multi_assay_experiment
  mae_data <- dataset("MAE", mae)
  data <- teal_data(mae_data)
  app <- init(
    data = data,
    modules = modules(
      module(
        label = "sampleVarSpec example",
        server = server,
        ui = ui,
        filters = "all"
      )
    )
  )
  shinyApp(app$ui, app$server)
}
if (interactive()) {
  my_app()
}