R/docker_app.R

In whtns/seuratTools: Tools for Managing Seurat Objects

#' Create a seurat app on a shinyproxy docker instance
#'
#' @param object a seurat object
#' @param loom_path path to a loom file
#' @param appTitle
#' @param organism_type
#' @param futureMb
#' #' @param bigwigb  path to sqlite database listing bigwig files for cells in the seurat object
#'
#' @return
#' @export
#'
#' @examples
#' \dontrun{
#' dockerSeuratApp(panc8)
#' }
#'
dockerSeuratApp <- function(object = panc8, loom_path = NULL, appTitle = NULL,
    organism_type = "human", futureMb = 13000, bigwig_db = "~/.cache/seuratTools/bw-files.db") {
    future::plan(strategy = "multicore", workers = 6)
    future_size <- futureMb * 1024^2
    options(future.globals.maxSize = future_size)
    options(shiny.maxRequestSize = 40 * 1024^2)
    options(DT.options = list(
        pageLength = 2000, paging = FALSE,
        info = TRUE, searching = TRUE, autoWidth = F, ordering = TRUE,
        scrollX = TRUE, language = list(search = "Filter:")
    ))
    header <- shinydashboard::dashboardHeader(title = appTitle)
    sidebar <- shinydashboard::dashboardSidebar(
        textOutput("appTitle"),
        uiOutput("featureType"),
        shinydashboard::sidebarMenu(
            shinydashboard::menuItem("Reformat Metadata",
                tabName = "reformatMetadata", icon = icon("columns")
            ), shinydashboard::menuItem("Plot Data",
                tabName = "comparePlots", icon = icon("chart-bar"), selected = TRUE
            ), shinydashboard::menuItem("Heatmap/Violin Plots",
                tabName = "violinPlots", icon = icon("sort")
            ), shinydashboard::menuItem("Coverage Plots",
                tabName = "coveragePlots", icon = icon("mountain")
            ), shinydashboard::menuItem("Differential Expression",
                tabName = "diffex", icon = icon("magnet")
                # ), shinydashboard::menuItem("Pathway Enrichment Analysis",
                #   tabName = "pathwayEnrichment", icon = icon("sitemap")
            ), shinydashboard::menuItem("Find Markers",
                tabName = "findMarkers", icon = icon("bullhorn")
            ), shinydashboard::menuItem("Subset Seurat Input",
                tabName = "subsetSeurat", icon = icon("filter")
            ), shinydashboard::menuItem("All Transcripts",
                tabName = "allTranscripts", icon = icon("sliders-h")
            ), shinydashboard::menuItem("Monocle",
                tabName = "monocle", icon = icon("bullseye")
            ), shinydashboard::menuItem("Regress Features",
                tabName = "regressFeatures", icon = icon("eraser")
            ), shinydashboard::menuItem("Technical Information",
                tabName = "techInfo", icon = icon("cogs")
            )
        ),
        actionButton("changeEmbedAction",
            label = "Change Embedding Parameters"
        ),
        changeEmbedParamsui("changeembed"),
        width = 250
    )
    body <- shinydashboard::dashboardBody(
        waiter::use_waiter(),
        shinydashboard::tabItems(
            shinydashboard::tabItem(
                tabName = "comparePlots",
                h2("Compare Plots") %>%
                    default_helper(type = "markdown", content = "comparePlots"),
                plotDimRedui("plotdimred1"),
                plotDimRedui("plotdimred2"),
                plotReadCountui("plotreadcount1"),
                plotReadCountui("plotreadcount2"),
                seuratToolsBox(
                    title = "Selected Cells",
                    tableSelectedui("tableselected"),
                    width = 6
                ),
                plotClustree_UI("clustreePlot")
            ),
            shinydashboard::tabItem(
                tabName = "violinPlots",
                fluidRow(
                    plotHeatmapui("heatMap")
                ),
                fluidRow(
                    plotViolinui("violinPlot")
                )
            ),
            shinydashboard::tabItem(
                tabName = "coveragePlots",
                fluidRow(
                    plotCoverage_UI("coverageplots")
                )
            ),
            shinydashboard::tabItem(
                tabName = "reformatMetadata",
                fluidRow(
                    reformatMetadataDRui("reformatMetadataDR")
                )
            ),
            shinydashboard::tabItem(
                tabName = "subsetSeurat",
                h2("Subset Seurat Input") %>%
                    default_helper(type = "markdown", content = "subsetSeurat"),
                plotDimRedui("subset"),
                seuratToolsBox(
                    title = "Subset Settings",
                    checkboxInput("legacySettingsSubset", "Use Legacy Settings", value = FALSE),
                    actionButton("subsetAction", "subset seurat by selected cells"),
                    actionButton("subsetCsv", "subset seurat by uploaded csv"),
                    fileInput("uploadCsv",
                        "Upload .csv file with cells to include",
                        accept = c(".csv")
                    ),
                    shinyjs::useShinyjs(),
                    # shinyjs::runcodeUI(code = "shinyjs::alert('Hello!')"),
                    textOutput("subsetMessages"),
                    width = 6
                ),
                seuratToolsBox(
                    title = "Selected Cells", tableSelectedui("subset"),
                    width = 6
                )
            ), shinydashboard::tabItem(
                tabName = "findMarkers",
                h2("Find Markers"),
                findMarkersui("findmarkers"),
                plotDimRedui("markerScatter")
            ), shinydashboard::tabItem(
                tabName = "allTranscripts",
                h2("All Transcripts"),
                plotDimRedui("alltranscripts2"),
                allTranscriptsui("alltranscripts1")
            ),
            shinydashboard::tabItem(
                tabName = "diffex",
                h2("Differential Expression") %>%
                    default_helper(type = "markdown", content = "diffex"),
                plotDimRedui("diffex"),
                diffexui("diffex")
            ),
            shinydashboard::tabItem(
                tabName = "pathwayEnrichment",
                h2("Pathway Enrichment"),
                fluidRow(
                    pathwayEnrichmentui("pathwayEnrichment")
                )
            ),
            shinydashboard::tabItem(
                tabName = "regressFeatures",
                fluidRow(
                    seuratToolsBox(
                        title = "Regress Features",
                        actionButton(
                            "regressAction",
                            "Regress Seurat Objects By Genes"
                        ),
                        checkboxInput("runRegression",
                            "Run Regression?",
                            value = FALSE
                        ), radioButtons("priorGeneSet",
                            "Choose a marker gene set:",
                            choices = c(
                                "Apoptosis",
                                "Cell Cycle",
                                "Mitochondrial",
                                "Ribosomal"
                            )
                        ), selectizeInput("geneSet",
                            "List of genes",
                            choices = NULL, multiple = TRUE
                        ),
                        textInput("geneSetName", "Name for Gene Set"),
                        width = 12
                    ) %>%
                        default_helper(type = "markdown", content = "regressFeatures")
                )
            ), shinydashboard::tabItem(
                tabName = "monocle",
                h2("Monocle"),
                monocleui("monocle")
            ), shinydashboard::tabItem(
                tabName = "techInfo",
                h2("Technical Information"),
                techInfoui("techInfo")
            )
        )
    )

    ui <- function(request) {
        ui <- dashboardPage(
            header = header, sidebar = sidebar,
            body = body
        )
    }
    server <- function(input, output, session) {
        w <- waiter::Waiter$new()

        shinyhelper::observe_helpers(help_dir = system.file("helpers", package = "seuratTools", lib.loc = "/dataVolume/storage/rpkgs/devel_install/"))
        options(warn = -1)
        # shinylogs::track_usage(storage_mode = shinylogs::store_json(path = "logs/"))
        # projects_db <- "/dataVolume/storage/single_cell_projects/single_cell_projects.db"

        seu <- reactiveVal(NULL)
        observe({
            seu(object)
        })

        organism_type <- reactive({
            "human"
        })

        loom_path <- reactive({
            loom_path
        })

        plot_types <- reactive({
            req(seu())
            list_plot_types(seu())
        })

        featureType <- reactive({
            "gene"
            # input$feature_type
        })


        observe({
            reformatted_seu <- callModule(reformatMetadataDR, "reformatMetadataDR", seu, featureType)
            seu(reformatted_seu())
        })

        reductions <- reactive({
            req(seu())
            names(seu()@reductions)
            # c("pca", "tsne", "umap")
        })

        observe({
            req(seu())

            callModule(plotDimRed, "plotdimred1", seu, plot_types, featureType,
                organism_type = organism_type, reductions
            )
            callModule(plotDimRed, "plotdimred2", seu, plot_types, featureType,
                organism_type = organism_type, reductions
            )
            callModule(plotDimRed, "diffex", seu, plot_types, featureType,
                organism_type = organism_type, reductions
            )
            callModule(plotDimRed, "subset", seu, plot_types, featureType,
                organism_type = organism_type, reductions
            )
            callModule(plotDimRed, "markerScatter", seu, plot_types, featureType,
                organism_type = organism_type, reductions
            )
        })

        callModule(plotReadCount, "plotreadcount1", seu, plot_types)
        callModule(plotReadCount, "plotreadcount2", seu, plot_types)

        callModule(
            plotViolin, "violinPlot", seu, featureType,
            organism_type
        )
        callModule(
            plotHeatmap, "heatMap", seu, featureType,
            organism_type
        )
        callModule(
            plotCoverage, "coverageplots", seu, plot_types, proj_dir, organism_type, bigwig_db
        )
        callModule(plotClustree, "clustreePlot", seu)
        callModule(tableSelected, "tableselected", seu)
        diffex_selected_cells <- callModule(
            tableSelected, "diffex",
            seu
        )
        subset_selected_cells <- callModule(
            tableSelected, "subset",
            seu
        )

        observeEvent(input$changeEmbedAction, {
            showModal(modalDialog(
                title = "Recalculating Embedding",
                "This process may take a minute or two!"
            ))
            seu <- callModule(
                changeEmbedParams, "changeembed",
                seu
            )
            removeModal()
        })

        callModule(findMarkers, "findmarkers", seu, plot_types, featureType)

        callModule(pathwayEnrichment, "pathwayEnrichment", seu)

        # plot all transcripts
        observe({
            req(featureType())
            req(seu())
            if ("transcript" %in% names(seu()@assays)) {
                callModule(
                    allTranscripts, "alltranscripts1", seu, featureType,
                    organism_type
                )

                callModule(plotDimRed, "alltranscripts2", seu, plot_types, featureType,
                    organism_type = organism_type, reductions
                )
            }
        })

        diffex_results <- callModule(
            diffex, "diffex", seu, featureType,
            diffex_selected_cells
        )
        observeEvent(input$subsetAction, {
            req(input$subsetAction)
            req(subset_selected_cells())
            withCallingHandlers(
                {
                    shinyjs::html("subsetMessages", "")
                    message("Beginning")
                    subset_seu <- seu()[, colnames(seu()) %in% subset_selected_cells()]
                    seu(subset_seu)
                    if (length(unique(seu()$batch)) > 1) {
                        message(paste0("reintegrating gene expression"))
                        reintegrated_seu <- reintegrate_seu(seu(),
                            resolution = seq(0.2, 2, by = 0.2),
                            organism = seu()@misc$experiment$organism
                        )
                        seu(reintegrated_seu)
                    } else {
                        processed_seu <- seurat_process(seu(), resolution = seq(0.2, 2, by = 0.2))
                        seu(processed_seu)
                    }
                    message("Complete!")
                },
                message = function(m) {
                    shinyjs::html(id = "subsetMessages", html = paste0(
                        "Subsetting Seurat Object: ",
                        m$message
                    ), add = FALSE)
                }
            )
        })
        observeEvent(input$subsetCsv, {
            req(input$subsetCsv)
            req(input$uploadCsv)
            withCallingHandlers(
                {
                    shinyjs::html("subsetMessages", "")
                    message("Beginning")
                    subset_seu <- subset_by_meta(
                        input$uploadCsv$datapath,
                        seu()
                    )
                    seu(subset_seu)

                    if (length(unique(seu()[[]]$batch)) > 1) {
                        message(paste0("reintegrating gene expression"))
                        reintegrated_seu <- reintegrate_seu(seu(),
                            resolution = seq(0.2, 2, by = 0.2),
                            organism = seu()@misc$experiment$organism
                        )
                        seu(reintegrated_seu)
                    } else {
                        processed_seu <- seurat_process(seu(), resolution = seq(0.2, 2, by = 0.2))
                        seu(processed_seu)
                    }
                    message("Complete!")
                },
                message = function(m) {
                    shinyjs::html(id = "subsetMessages", html = paste0(
                        "Subsetting Seurat Object: ",
                        m$message
                    ), add = FALSE)
                }
            )
        })

        prior_gene_set <- reactive({
            # req(input$priorGeneSet)
            req(seu())

            if (is.null(input$priorGeneSet)) {
                ""
            } else if (input$priorGeneSet == "Apoptosis") {
                marker_genes <- c(
                    "CASP3", "CASP7", "BAX", "BAK1", "BID", "BBC3",
                    "BCL2", "MCL1"
                )

                marker_genes[marker_genes %in% rownames(seu())]
            } else if (input$priorGeneSet == "Cell Cycle") {
                marker_genes <- c(
                    "MCM5", "PCNA", "TYMS", "FEN1", "MCM2", "MCM4",
                    "RRM1", "UNG", "GINS2", "MCM6", "CDCA7", "DTL",
                    "PRIM1", "UHRF1", "MLF1IP", "HELLS", "RFC2",
                    "RPA2", "NASP", "RAD51AP1", "GMNN", "WDR76",
                    "SLBP", "CCNE2", "UBR7", "POLD3", "MSH2",
                    "ATAD2", "RAD51", "RRM2", "CDC45", "CDC6",
                    "EXO1", "TIPIN", "DSCC1", "BLM", "CASP8AP2",
                    "USP1", "CLSPN", "POLA1", "CHAF1B", "BRIP1",
                    "E2F8"
                )

                marker_genes[marker_genes %in% rownames(seu())]
            } else if (input$priorGeneSet == "Mitochondrial") {
                marker_genes <- mito_features[[organism_type()]][["gene"]]

                marker_genes[marker_genes %in% rownames(seu())]
            } else if (input$priorGeneSet == "Ribosomal") {
                marker_genes <- ribo_features[[organism_type()]][["gene"]]

                marker_genes[marker_genes %in% rownames(seu())]
            }
        })

        observe({
            req(seu())
            updateSelectizeInput(session, "geneSet",
                choices = rownames(seu()),
                selected = prior_gene_set(),
                server = TRUE
            )
        })
        observeEvent(input$regressAction, {
            req(seu())
            showModal(modalDialog(
                title = "Regressing out provided list of features",
                "This process may take a minute or two!"
            ))
            regressed_seu <- seuratTools::regress_by_features(seu(),
                feature_set = list(input$geneSet), set_name = janitor::make_clean_names(input$geneSetName),
                regress = input$runRegression
            )
            seu(regressed_seu)
            removeModal()
        })

        observe({
            req(reductions())
            callModule(
                monocle, "monocle", seu, plot_types, featureType,
                organism_type, reductions
            )
        })

        callModule(techInfo, "techInfo", seu)
    }
    shinyApp(ui, server, enableBookmarking = "server")
}