Introduction

teal.modules.hermes is a package implementing a number of teal modules for the exploration of RNA-sequencing counts data. In addition to predefined modules, teal.modules.hermes enables quick and easy ad-hoc module creation.

Ad-hoc module example

Let’s assume you have a function awesome_plot() which takes a count matrix and makes an awesome plot out of it. Now you would like to make a Shiny app where you can filter patients, samples, select the experiment out of your MultiAssayExperiment (MAE), select the count matrix from the experiment, etc. Nothing is easier than that with teal.modules.hermes! We show you below how to quickly spin up your UI, server and put them together into a nice little app.

UI function

In teal.modules.hermes we provide modules that make the experiment and assay selection super easy, see here for the UI part:

ui <- function(id, mae_name) {
  ns <- NS(id)

  teal.widgets::standard_layout(
    encoding = uiOutput(ns("encoding_ui")),
    output = plotOutput(ns("awesome_plot"))
  )
}

Server function

Similarly for the server we use the modules, and call then our awesome plotting function.

srv <- function(input,
                output,
                session,
                data,
                filter_panel_api,
                mae_name) {
  output$encoding_ui <- renderUI({
    tags$div(
      experimentSpecInput(session$ns("experiment"), data, mae_name),
      assaySpecInput(session$ns("assay"))
    )
  })
  experiment <- experimentSpecServer(
    "experiment",
    data = data,
    filter_panel_api = filter_panel_api,
    mae_name = mae_name,
    name_annotation = NULL # If you have a gene name column in your rowData, can specify here.
  )
  assay <- assaySpecServer("assay", experiment$assays)
  output$awesome_plot <- renderPlot({
    data <- experiment$data()
    assay <- assay()
    req(assay %in% SummarizedExperiment::assayNames(data))
    counts <- SummarizedExperiment::assay(data, assay)
    df <- data.frame(gene = rownames(counts), counts = rowSums(counts))
    df <- na.omit(df[order(df$counts, decreasing = TRUE), ])
    df$gene <- factor(df$gene, levels = df$gene)
    df <- df[1:10, ]
    ggplot(df, aes(x = gene, y = counts)) +
      geom_col() +
      theme(axis.text.x = element_text(angle = 90))
  })
}

App function

Now let’s assume you want to spin up your app for an MAE.

awesome_app <- function(mae, label = "My awesome app") {
  mae_name <- "MAE"
  data <- teal_data(MAE = hermes::lapply(mae, hermes::HermesData))
  app <- init(
    data = data,
    modules = teal::modules(
      module(
        label = label,
        server = srv,
        server_args = list(mae_name = mae_name),
        ui = ui,
        ui_args = list(mae_name = mae_name),
        datanames = mae_name
      )
    )
  )
  shinyApp(app$ui, app$server)
}

Testing it

To test this:

awesome_app(hermes::multi_assay_experiment)

- Introduction
- Ad-hoc module example