R/shiny.R
In markrobinsonuzh/iCOBRA: Comparison and Visualization of Ranking and Assignment Methods
Defines functions
COBRAapp
Documented in
COBRAapp
#' Interactive shiny app to visualize results
#'
#' Interactive shiny app for visualization of results. The app can be
#' initialized with a \code{COBRAData} object. If no object is provided, truth
#' and results are loaded into the app from text files (see the Instructions tab
#' of the app for formatting instructions). Properly formatted text files can
#' also be obtained using the function \code{\link{COBRAData_to_text}}.
#'
#' @param cobradata An (optional) \code{COBRAData} object. If not given, the
#' user can load results from text files.
#' @param autorun A logical indicating whether the app calculations should start
#' automatically on launch, or wait for the user to press the 'Start
#' calculation!' button.
#' @author Charlotte Soneson
#' @return Returns (and runs) an object representing the shiny app.
#' @import shiny
#' @import shinydashboard
#' @importFrom shinyBS bsTooltip
#' @importFrom utils packageVersion
#' @export
#' @examples
#' data(cobradata_example)
#' app <- COBRAapp(cobradata_example)
#' if (interactive()) {
#' shiny::runApp(app)
#' }
COBRAapp <- function(cobradata = NULL, autorun = FALSE) {
## ------------------------------------------------------------------ ##
## Define UI ##
## ------------------------------------------------------------------ ##
p_layout <-
shinydashboard::dashboardPage(
skin = "blue",
shinydashboard::dashboardHeader(
title = paste0("iCOBRA - interactive COmparative evaluation of ",
"Binary classification and RAnking methods (v",
packageVersion("iCOBRA"), ")"),
titleWidth = 950),
shinydashboard::dashboardSidebar(
width = 350,
## Settings and inputs for the truth
shinydashboard::menuItem("Truth", icon = icon("database"),
## Load the file containing the truth.
uiOutput("choose_truth_file"),
shinyBS::bsTooltip(
"truth", paste0("Select the file containing the true status",
" of each feature. See the Instructions ",
"tab for formatting instructions."),
"right", options = list(container = "body")),
## Define the column containing the feature identifier
## (for both truth and results)
uiOutput("choose_feature_id"),
## Define the column containing the (binary) truth
uiOutput("choose_binary_truth"),
## Define the column containing the (continuous) truth
uiOutput("choose_continuous_truth"),
## Define the variable used to stratify the results,
## based on the columns available in the truth file.
uiOutput("splitvar"),
shinyBS::bsTooltip(
"splitvar",
paste0("Select a feature attribute by which to stratify ",
"the result representations."),
"right", options = list(container = "body")),
## Define the maximal number of categories to retain
## in stratification.
uiOutput("choosemaxsplit"),
shinyBS::bsTooltip(
"choosemaxsplit",
paste0("Set the number of categories to show if the results",
" are stratified by a variable annotation. The most ",
"frequent categories with both positive and negative",
" instances will be retained"),
"right", options = list(container = "body")),
## Decide whether or not to include the "overall" category when
## stratifying the results.
uiOutput("chooseincludeoverall"),
shinyBS::bsTooltip(
"includeoverall",
paste0("Select whether or not to include the 'overall' ",
"class when showing stratified results."),
"right", options = list(container = "body"))),
## Settings and inputs for results
shinydashboard::menuItem("Results", icon = icon("folder-open"),
## Load the file containing the results.
uiOutput("choose_result_file"),
shinyBS::bsTooltip(
"file1", paste0("Select file containing results from one or",
" multiple methods. See the Instructions ",
"tab for formatting instructions."),
"right", options = list(container = "body")),
## Decide which methods to include in the results.
## Depends on the loaded result files.
uiOutput("columns"),
shinyBS::bsTooltip(
"columns",
paste0("Select the methods for which to show the results."),
"right", options = list(container = "body")),
## Decide whether to base evaluations on all genes in
## truth table, or only on variables for which both a truth
## value and a result is provided
checkboxInput(
inputId = "onlyshared",
label = paste0("Calculate performance based only on ",
"features shared between truth and result ",
"tables."),
value = FALSE),
shinyBS::bsTooltip(
"onlyshared",
paste0("Calculate performance based only on ",
"features shared between truth and result ",
"tables. Otherwise, all features in the ",
"truth table will be used."),
"right", options = list(container = "body")),
actionButton("goButton", "Start calculation!",
icon = icon("plane"))),
## Plot settings
shinydashboard::menuItem("Plot settings", icon = icon("paint-brush"),
## Choose color scheme
selectInput(
"colorscheme", "Select color palette",
c("hue_pal", "Accent (max 7 methods)",
"Dark2 (max 7 methods)", "Paired (max 11 methods)",
"Pastel1 (max 8 methods)", "Pastel2 (max 7 methods)",
"Set1 (max 8 methods)", "Set2 (max 7 methods)",
"Set3 (max 11 methods)", "rainbow", "heat", "terrain",
"topo", "cm"),
selectize = TRUE),
shinyBS::bsTooltip(
"colorscheme",
paste0("Choose color palette. Some palettes are only ",
"applicable if the number of methods ",
"(or method/stratification level combinations) is ",
"below a certain threshold. If this threshold is ",
"exceeded, the colorscheme will default to hue_pal."),
"right", options = list(container = "body")),
## Decide what to include in FDR/TPR plots.
checkboxGroupInput(
"plottype",
paste0("Display full curve and/or points in FDR/TPR, ",
"FDR/NBR and FSR/NBR plots"),
c("curve", "points"), c("points")),
## Decide whether or not to facet plots.
checkboxGroupInput(
"facet_opt",
"Split plots into panels by stratifying variable",
"split", "split"),
## Define the q-value thresholds to use in the plots.
textInput(inputId = "fdrthresholds", label = "FDR thresholds",
value = "0.01, 0.05, 0.1"),
shinyBS::bsTooltip("fdrthresholds",
paste0("Specific FDR thresholds at which the ",
"performance will be evaluated. ",
"Separate multiple values with comma"),
"right", options = list(container = "body")),
## Define the s-value thresholds to use in the plots.
textInput(inputId = "svalthresholds",
label = "s-value thresholds",
value = "0.01, 0.05, 0.1"),
shinyBS::bsTooltip("svalthresholds",
paste0("Specific s-value thresholds at ",
"which the performance will be ",
"evaluated. Separate multiple ",
"values with comma"),
"right",
options = list(container = "body")),
## Define the plot height (in pixels).
numericInput(inputId = "plotheight",
label = "Plot height (numeric, in pixels)",
value = 600, min = 200, max = 2000, step = 10),
shinyBS::bsTooltip("plotheight",
paste0("The height of the plots (in pixels). ",
"Default 800."),
"right", options = list(container = "body")),
## Define the pointsize used in the plots.
numericInput(inputId = "pointsize", label = "Point size",
value = 5),
shinyBS::bsTooltip("pointsize",
paste0("The point size used in the plots."),
"right", options = list(container = "body")),
## Define the fontsize used in the panel headers.
numericInput(inputId = "stripsize",
label = "Font size for panel headers",
value = 15),
shinyBS::bsTooltip("stripsize",
paste0("The font size used for panel headers in",
" facetted plots."),
"right", options = list(container = "body"))),
## Button to close app
shiny::actionButton("close_app", "Close app")
),
## Outputs
shinydashboard::dashboardBody(fluidRow(
shinydashboard::tabBox(
width = 12,
## Define output size of error messages
tags$head(
tags$style(HTML("
.shiny-output-error-validation {
font-size: 17px;
}
"))
),
tabPanel("About",
includeMarkdown(system.file("extdata", "about.md",
package = "iCOBRA")),
value = "about"),
tabPanel("Instructions",
includeMarkdown(system.file("extdata", "instructions.md",
package = "iCOBRA")),
value = "instructions"),
tabPanel("Data preview",
h1("Truth"),
DT::dataTableOutput("table.truth"),
h1("p-values"),
DT::dataTableOutput("table.pval"),
h1("adjusted p-values"),
DT::dataTableOutput("table.padj"),
h1("s-values"),
DT::dataTableOutput("table.sval"),
h1("scores"),
DT::dataTableOutput("table.score"),
value = "datapreview"),
tabPanel("TPR vs FDR",
uiOutput("plot.fdrtprcurve"),
fluidRow(
column(3, sliderInput(inputId = "xrange_fdrtpr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(3, sliderInput(inputId = "yrange_fdrtpr",
label = "y-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fdrtprcurve",
label = "Download plot")),
column(2, br(), downloadButton("export.fdrtpr.df.rdata",
label = "Download Rdata")),
column(2, br(), uiOutput("fdrtpr.df.tsv.button"))
),
DT::dataTableOutput("fdrtprcurve_click_info"),
value = "fdrtprcurve"),
tabPanel("NBR vs FDR",
uiOutput("plot.fdrnbrcurve"),
fluidRow(
column(4, sliderInput(inputId = "xrange_fdrnbr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fdrnbrcurve",
label = "Download plot")),
column(2, br(), downloadButton("export.fdrnbr.df.rdata",
label = "Download Rdata")),
column(2, br(), uiOutput("fdrnbr.df.tsv.button"))
),
DT::dataTableOutput("fdrnbrcurve_click_info"),
value = "fdrnbrcurve"),
tabPanel("TPR", uiOutput("plot.tpr"),
fluidRow(
column(4, sliderInput(inputId = "xrange_tpr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.tpr",
label = "Download plot")),
column(2, br(), downloadButton("export.tpr.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.tpr.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("tpr_click_info"), value = "tpr"),
tabPanel("FPR", uiOutput("plot.fpr"),
fluidRow(
column(4, sliderInput(inputId = "xrange_fpr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fpr",
label = "Download plot")),
column(2, br(), downloadButton("export.fpr.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.fpr.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("fpr_click_info"), value = "fpr"),
tabPanel("ROC", uiOutput("plot.roc"),
fluidRow(
column(3, sliderInput(inputId = "xrange_roc",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(3, sliderInput(inputId = "yrange_roc",
label = "y-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.roc",
label = "Download plot")),
column(2, br(), downloadButton("export.roc.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.roc.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("roc_click_info"), value = "roc"),
tabPanel("False discovery curves",
uiOutput("plot.fpc"),
fluidRow(
column(3, numericInput(inputId = "maxnfdc",
label = "Maximal rank to display",
value = 500, min = 10, max = 10000,
step = 10)),
column(2, br(), downloadButton("export.fpc",
label = "Download plot")),
column(2, br(), downloadButton("export.fpc.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.fpc.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("fpc_click_info"), value = "fpc"),
tabPanel("Correlation",
uiOutput("plot.corr"),
fluidRow(
column(3, sliderInput(inputId = "xrange_corr",
label = "x-axis limits",
min = -1, max = 1, value = c(-1, 1),
step = 0.01)),
column(2, radioButtons(inputId = "corrtype",
label = "Correlation measure",
choices = c("pearson", "spearman"),
selected = "spearman")),
column(2, br(), downloadButton("export.corr",
label = "Download plot")),
column(2, br(), downloadButton("export.corr.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.corr.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("corr_click_info"), value = "corr"),
tabPanel("Scatter",
uiOutput("plot.scatter"),
fluidRow(
column(2, checkboxInput("doflip", "Flip axes", FALSE),
checkboxInput("dolog", "Log-transform", FALSE)),
column(2, br(), downloadButton("export.scatter",
label = "Download plot")),
column(2, br(), downloadButton("export.scatter.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.scatter.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("scatter_click_info"),
value = "scatter"),
tabPanel("Deviations",
uiOutput("plot.deviation"),
fluidRow(
column(2, checkboxInput("dojitter",
"Include jittered points", FALSE),
radioButtons(inputId = "devtrans",
label = "Transformation",
choices = c("raw", "absolute",
"squared"),
selected = "raw")),
column(1, radioButtons(
inputId = "devtype", label = "Plot type",
choices = c("boxplot", "violinplot"),
selected = "violinplot")),
column(3, uiOutput("axislimitsdeviation")),
column(2, br(), downloadButton("export.deviation",
label = "Download plot")),
column(2, br(), downloadButton("export.deviation.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.deviation.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("deviation_click_info"),
value = "deviation"),
tabPanel("Venn diagram",
uiOutput("plot.overlap"),
fluidRow(
column(1, radioButtons(inputId = "incltruth",
label = "Include truth",
choices = c("yes", "no"),
selected = "yes")),
shinyBS::bsTooltip(
"incltruth",
paste0("Whether or nor to include the truth as a ",
"(perfect) method in the Venn diagrams. Note ",
"that maximally five methods (including the ",
"truth) can be included."),
"bottom", options = list(container = "body")),
column(1, radioButtons(inputId = "vennType",
label = "Type of threshold",
choices = c("adjp", "rank"),
selected = "adjp")),
shinyBS::bsTooltip(
"vennType",
paste0("How to determine the set of features to ",
"compare between methods. "),
"bottom", options = list(container = "body")),
column(2, numericInput(
inputId = "adjpVenn",
label = "Adjusted p-value threshold",
value = 0.05, min = 0, max = 1, step = 0.01)),
shinyBS::bsTooltip(
"adjpVenn",
paste0("The adjusted p-value threshold used to extract ",
"the sets of significant variables to use for ",
"the Venn diagram. "),
"bottom", options = list(container = "body")),
column(2, numericInput(
inputId = "rankVenn",
label = "Rank threshold",
value = 100, min = 0, max = 1e10, step = 1)),
shinyBS::bsTooltip(
"rankVenn",
paste0("The rank used to extract the top-ranked ",
"features to compare in the Venn diagram. "),
"bottom", options = list(container = "body")),
column(2, br(), downloadButton("export.overlap",
label = "Download plot")),
column(2, br(), downloadButton("export.overlap.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.overlap.df.tsv",
label = "Download tsv"))
),
value = "overlap"),
tabPanel("UpSet plot",
uiOutput("plot.upset"),
fluidRow(
column(2, uiOutput("upset.stratum")),
column(2, radioButtons(inputId = "order.upset.by",
label = "Order intersections by",
choices = c("degree", "freq"),
selected = "freq")),
column(2, radioButtons(inputId = "upset.order",
label = "Order intersections",
choices = c("decreasing",
"increasing"),
selected = "decreasing")),
column(2, br(), downloadButton("export.upset",
label = "Download plot")),
column(2, br(), downloadButton("export.upset.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.upset.df.tsv",
label = "Download tsv"))
),
value = "upset"),
tabPanel("NBR vs FSR",
uiOutput("plot.fsrnbrcurve"),
fluidRow(
column(4, sliderInput(inputId = "xrange_fsrnbr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fsrnbrcurve",
label = "Download plot")),
column(2, br(), downloadButton("export.fsrnbr.df.rdata",
label = "Download Rdata")),
column(2, br(), uiOutput("fsrnbr.df.tsv.button"))
),
DT::dataTableOutput("fsrnbrcurve_click_info"),
value = "fsrnbrcurve"),
selected = "fdrtprcurve"
)
))
)
options(shiny.maxRequestSize = 15*1024^2)
## ------------------------------------------------------------------ ##
## Define server ##
## ------------------------------------------------------------------ ##
#nocov start
server_function <- function(input, output, session) {
values <- reactiveValues()
values$my_cobradata <- COBRAData()
output$choose_truth_file <- renderUI({
if (!is.null(cobradata)) {
NULL
} else {
return(fileInput(inputId = "truth", label = "Load truth file",
accept = c("text/csv", "text/comma-separated-values",
"text/tab-separated-values", "text/plain",
".csv", ".tsv", ".tab", ".txt"),
multiple = FALSE))
}
})
truthFile <- reactive({
if (is.null(input$truth))
return(NULL)
trf <- utils::read.delim(input$truth$datapath, header = TRUE,
as.is = TRUE, sep = "\t", quote = "",
check.names = FALSE)
isolate(values$my_cobradata <-
COBRAData(truth = trf,
object_to_extend = values$my_cobradata))
return(trf)
})
## Render the UI element to choose the column containing the feature ID
output$choose_feature_id <- renderUI({
if (is.null(truthFile()))
return(NULL)
selectInput("feature_id", "Select column containing feature identifiers",
colnames(truthFile()), selectize = TRUE)
})
## Render the UI element to choose the binary truth column
output$choose_binary_truth <- renderUI({
if (!is.null(cobradata)) {
ccl <- sapply(truth(cobradata), class)
only01 <- sapply(truth(cobradata), function(i) {
if (all(i %in% c(0, 1, NA))) "binary"
else "notbinary"
})
} else {
if (is.null(truthFile()))
return(NULL)
ccl <- sapply(truthFile(), class)
only01 <- sapply(truthFile(), function(i) {
if (all(i %in% c(0, 1, NA))) "binary"
else "notbinary"
})
}
keepcols <- intersect(which(ccl %in% c("numeric", "integer")),
which(only01 == "binary"))
if (!is.null(cobradata)) {
selectInput("binary_truth", "Select column containing binary truth",
c(colnames(truth(cobradata))[keepcols], "none"),
selectize = TRUE)
} else {
selectInput("binary_truth", "Select column containing binary truth",
c(colnames(truthFile())[keepcols], "none"),
selectize = TRUE)
}
})
## Render the UI element to choose the continuous truth column
output$choose_continuous_truth <- renderUI({
if (!is.null(cobradata)) {
ccl <- sapply(truth(cobradata), class)
} else {
if (is.null(truthFile()))
return(NULL)
ccl <- sapply(truthFile(), class)
}
keepcols <- which(ccl %in% c("numeric", "integer"))
if (!is.null(cobradata)) {
selectInput("cont_truth", "Select column containing continuous truth",
c("none", colnames(truth(cobradata))[keepcols]),
selectize = TRUE)
} else {
selectInput("cont_truth", "Select column containing continuous truth",
c("none", colnames(truthFile())[keepcols]),
selectize = TRUE)
}
})
## Render the UI element to choose which attribute (from the truth file)
## to split panels by.
output$splitvar <- renderUI({
if (!is.null(cobradata)) {
selectInput("splv", "Select variable to stratify by",
c("none", colnames(truth(cobradata))), selectize = TRUE)
} else {
trf <- truthFile()
if (is.null(trf))
return(NULL)
selectInput("splv", "Select variable to stratify by",
c("none", colnames(trf)), selectize = TRUE)
}
})
## Render the UI element to choose how many categories to include.
output$choosemaxsplit <- renderUI({
numericInput(inputId = "maxsplit",
label = "Maximum number of levels to show when stratifying",
value = 3, min = 1)
})
## Render the UI element to determine whether the overall category should
## be included when stratifying
output$chooseincludeoverall <- renderUI({
radioButtons(inputId = "includeoverall",
label = "Include 'overall' class when stratifying",
choices = c("yes", "no"), selected = "yes")
})
## Render the UI element to upload result file(s)
output$choose_result_file <- renderUI({
if (!is.null(cobradata)) {
NULL
} else {
return(fileInput(inputId = "file1", label = "Add file with results",
accept = c("text/csv", "text/comma-separated-values",
"text/tab-separated-values", "text/plain",
".csv", ".tsv"), multiple = FALSE))
}
})
## Render the UI object for selecting methods to include, based on the
## uploaded result files.
output$columns <- renderUI({
if (!is.null(cobradata)) {
allm <- unique(c(colnames(pval(cobradata)), colnames(padj(cobradata)),
colnames(sval(cobradata)), colnames(score(cobradata))))
isolate(values$my_cobradata <- cobradata)
isolate(values$all_methods <- allm)
isolate(values$my_cobradata <- calculate_adjp(values$my_cobradata))
## Create the UI element for selecting methods to include.
checkboxGroupInput("cols", "Select methods",
isolate(values$all_methods),
isolate(values$all_methods))
} else {
rownames(truth(values$my_cobradata)) <-
truth(values$my_cobradata)[, input$feature_id]
## Read file
inFile <- input$file1
if (is.null(inFile))
return(NULL)
v <- utils::read.delim(inFile$datapath, header = TRUE, as.is = TRUE,
sep = "\t", quote = "", check.names = FALSE)
## Perform some checks on the file, if ok then continue
if (res_check(v)) {
## Extract all columns corresponding to methods
ii <- 1
while(ii <= ncol(v)) {
i <- colnames(v)[ii]
coltype <- get_coltype(i)
if (!(is.null(coltype))) {
## Fix duplicates
tmp1 <- v[, c(input$feature_id, i)]
tmp1 <- fix_duplicates(tmp1, input$feature_id, i)
## Rename column to only method name
tmp <- tmp1[i]
rownames(tmp) <- tmp1[, input$feature_id]
i <- gsub(":S", "", gsub(":score$", "", gsub(":adjP$", "",
gsub(":P$", "", i))))
colnames(tmp) <- i
## Add values to data collection
isolate(values$all_methods <- unique(c(values$all_methods, i)))
if (coltype == "pval") {
isolate(values$my_cobradata <-
COBRAData(pval = tmp,
object_to_extend = values$my_cobradata))
} else if (coltype == "padj") {
isolate(values$my_cobradata <-
COBRAData(padj = tmp,
object_to_extend = values$my_cobradata))
} else if (coltype == "sval") {
isolate(values$my_cobradata <-
COBRAData(sval = tmp,
object_to_extend = values$my_cobradata))
} else if (coltype == "score") {
isolate(values$my_cobradata <-
COBRAData(score = tmp,
object_to_extend = values$my_cobradata))
}
}
ii <- ii + 1
}
## Calculate adjusted p-values if needed
isolate(values$my_cobradata <- calculate_adjp(values$my_cobradata))
## Render the UI element
checkboxGroupInput("cols", "Select methods",
isolate(values$all_methods),
isolate(values$all_methods))
}
}
})
plotvalues_TPR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
thrs <- sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$fdrthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "tpr",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FDR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
thrs <- sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$fdrthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fdrtpr",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FPR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
thrs <- sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$fdrthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fpr",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_ROC <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "roc",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FPC <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fpc",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FDRTPR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fdrtprcurve",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FSR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
thrs <-
sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$svalthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "fsrnbr",
svalthrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FSRNBR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "fsrnbrcurve",
svalthrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_overlap <- reactive({
if ((input$goButton > 0 | isTRUE(autorun))) {
if (input$binary_truth == "none") {
bt <- NULL
} else {
bt <- input$binary_truth
}
return(calculate_performance(values$my_cobradata,
binary_truth = bt,
cont_truth = NULL,
aspects = "overlap",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = input$adjpVenn,
type_venn = input$vennType,
topn_venn = input$rankVenn))
} else {
return(COBRAPerformance())
}
})
plotvalues_corr <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "corr",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_scatter <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "scatter",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_deviation <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "deviation",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues <- reactive({
if ((input$goButton > 0 | isTRUE(autorun))) {
## Put together results for chosen methods
if (length(input$cols) != 0) {
withProgress(message = "Calculating...", value = 0, {
all_vals <-
COBRAPerformance(tpr = tpr(plotvalues_TPR()),
fdrtpr = fdrtpr(plotvalues_FDR()),
fdrnbr = fdrtpr(plotvalues_FDR()),
fdrtprcurve = fdrtprcurve(plotvalues_FDRTPR()),
fdrnbrcurve = fdrnbrcurve(plotvalues_FDRTPR()),
fsrnbr = fsrnbr(plotvalues_FSR()),
fsrnbrcurve = fsrnbrcurve(plotvalues_FSRNBR()),
fpr = fpr(plotvalues_FPR()),
roc = roc(plotvalues_ROC()),
fpc = fpc(plotvalues_FPC()),
overlap = overlap(plotvalues_overlap()),
maxsplit = input$maxsplit,
splv = input$splv,
corr = corr(plotvalues_corr()),
scatter = scatter(plotvalues_scatter()),
deviation = deviation(plotvalues_deviation()))
all_vals <- prepare_data_for_plot(
cobraperf = all_vals,
keepmethods = input$cols,
incloverall = (input$includeoverall == "yes"),
colorscheme = input$colorscheme,
facetted = ("split" %in% input$facet_opt),
incltruth = (input$incltruth == "yes"))
## This is a bit of a hack, just to make the plots directly
## depend on the truth input. Otherwise, they will not be
## automatically updated if the truth file is changed
## (not until something else that directly affects the plots
## is changed as well, like the FDR thresholds or plot size etc).
## The title is rendered in white, so it is not visible.
title <- input$truth$name
list(all_vals = all_vals, title = title)
})
}
} else {
NULL
}
})
## -------------------------- OVERLAP ------------------------------ ##
## Generate Venn diagram plot
output$plot.overlap <- renderUI({
plotOutput("overlap", width = "100%",
height = paste0(input$plotheight, "px"))
})
## Figure for exporting
output$export.overlap <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, height = 12, width = 12)
if (length(values$all_methods) == 0 | length(input$cols) == 0)
return(NULL)
plot_overlap(plotvalues()$all_vals)
grDevices::dev.off()
})
## Data for exporting
output$export.overlap.df.rdata <- downloadHandler(
filename = "overlap-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.overlap.df.tsv <- downloadHandler(
filename = "overlap-data.tsv",
content = function(file) {
overlap_df <- overlap(plotvalues()$all_vals)
if (is(overlap_df, "list")) {
overlap_df <- lapply(overlap_df,
function(w) cbind(feature = rownames(w), w))
for (i in seq_len(length(overlap_df))) {
overlap_df[[i]]$featureclass <- names(overlap_df)[i]
}
overlap_df <- do.call(rbind, overlap_df)
}
else {
overlap_df <- cbind(feature = rownames(overlap_df), overlap_df)
}
utils::write.table(overlap_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$overlap <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(is_plottable(overlap(plotvalues()$all_vals)),
paste0("Venn diagrams can not be displayed. Check that ",
"binary_truth is not 'none', and that adjusted ",
"p-values are provided."))
)
validate(
need((is(overlap(plotvalues()$all_vals), "list") &&
ncol(overlap(plotvalues()$all_vals)[[1]]) <= 5) |
(is(overlap(plotvalues()$all_vals), "data.frame") &&
ncol(overlap(plotvalues()$all_vals)) <= 5),
paste0("Venn diagrams can not be constructed if more than five ",
"methods are selected (including the truth, ",
"if applicable)."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun))) {
return(NULL)
}
plot_overlap(plotvalues()$all_vals)
})
})
## --------------------------- UPSET ------------------------------- ##
## Choose category
output$upset.stratum <- renderUI({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(overlap(plotvalues()$all_vals)))
return(NULL)
else {
if (!is(overlap(plotvalues()$all_vals), "list"))
selectInput("upset.stratum", "Select category",
"overall", selectize = TRUE)
else
selectInput("upset.stratum", "Select category",
names(overlap(plotvalues()$all_vals)),
selectize = TRUE)
}
})
## Generate UpSet plot
output$plot.upset <- renderUI({
plotOutput("upset", width = "100%",
height = paste0(input$plotheight, "px"))
})
## Figure for exporting
output$export.upset <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, height = 12, width = 12)
if (length(values$all_methods) == 0 | length(input$cols) == 0)
return(NULL)
plot_upset(plotvalues()$all_vals, order.by = input$order.upset.by,
decreasing = (input$upset.order == "decreasing"),
stratum = input$upset.stratum)
grDevices::dev.off()
})
## Data for exporting
output$export.upset.df.rdata <- downloadHandler(
filename = "upset-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.upset.df.tsv <- downloadHandler(
filename = "upset-data.tsv",
content = function(file) {
overlap_df <- overlap(plotvalues()$all_vals)
if (is(overlap_df, "list")) {
overlap_df <- lapply(overlap_df,
function(w) cbind(feature = rownames(w), w))
for (i in seq_len(length(overlap_df))) {
overlap_df[[i]]$featureclass <- names(overlap_df)[i]
}
overlap_df <- do.call(rbind, overlap_df)
}
else {
overlap_df <- cbind(feature = rownames(overlap_df), overlap_df)
}
utils::write.table(overlap_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$upset <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(is_plottable(overlap(plotvalues()$all_vals)),
paste0("UpSet plots can not be displayed. Check that ",
"binary_truth is not 'none', and that adjusted ",
"p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun))) {
return(NULL)
}
plot_upset(plotvalues()$all_vals, order.by = input$order.upset.by,
decreasing = (input$upset.order == "decreasing"),
stratum = input$upset.stratum)
})
})
## ---------------------------- TPR -------------------------------- ##
output$plot.tpr <- renderUI({
plotOutput("tpr", width = "100%", height = paste0(input$plotheight, "px"),
hover = "tpr_plot_click")
})
## Figure for exporting
output$export.tpr <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(tpr(plotvalues()$all_vals)))
return(NULL)
print(plot_tpr(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_tpr))
grDevices::dev.off()
})
## Data for exporting
output$export.tpr.df.rdata <- downloadHandler(
filename = "tpr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.tpr.df.tsv <- downloadHandler(
filename = "tpr-data.tsv",
content = function(file) {
tpr_df <- isolate(tpr(plotvalues()$all_vals))
utils::write.table(tpr_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$tpr <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(tpr(plotvalues()$all_vals)),
paste0("TPR can not be displayed. Check that ",
"binary_truth is not 'none' and that adjusted ",
"p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(tpr(plotvalues()$all_vals))) {
return(NULL)
}
plot_tpr(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, xaxisrange = input$xrange_tpr)
})
})
output$tpr_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(tpr(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(tpr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$tpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method", xvar = "TPR")
} else {
res <- nearPoints(all_data, input$tpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, yvar = "num_method", xvar = "TPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("TPR"),
tabtype = "large")
})
## ------------------------ DATA PREVIEW --------------------------- ##
output$table.truth <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(truth(values$my_cobradata))
})
output$table.pval <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(pval(values$my_cobradata))
})
output$table.padj <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(padj(values$my_cobradata))
})
output$table.sval <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(sval(values$my_cobradata))
})
output$table.score <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(score(values$my_cobradata))
})
## ---------------------------- CORR ------------------------------- ##
output$plot.corr <- renderUI({
plotOutput("corr", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "corr_plot_click")
})
## Figure for exporting
output$export.corr <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(corr(plotvalues()$all_vals)))
return(NULL)
print(plot_corr(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_corr,
corrtype = input$corrtype))
grDevices::dev.off()
})
## Data for exporting
output$export.corr.df.rdata <- downloadHandler(
filename = "corr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.corr.df.tsv <- downloadHandler(
filename = "corr-data.tsv",
content = function(file) {
tpr_df <- isolate(corr(plotvalues()$all_vals))
utils::write.table(tpr_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$corr <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(corr(plotvalues()$all_vals)),
paste0("Correlation plots can not be displayed. Check that ",
"cont_truth is not 'none' and that scores ",
"are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(corr(plotvalues()$all_vals)))
return(NULL)
plot_corr(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, xaxisrange = input$xrange_corr,
corrtype = input$corrtype)
})
})
output$corr_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(corr(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(corr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$corr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method", xvar = toupper(input$corrtype))
} else {
res <- nearPoints(all_data, input$corr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, yvar = "num_method",
xvar = toupper(input$corrtype))
}
fix_res(res, methodcol = "fullmethod", aspcts = toupper(input$corrtype),
tabtype = "corr")
})
## -------------------------- DEVIATION ---------------------------- ##
output$axislimitsdeviation <- renderUI({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
else {
if (input$devtrans == "raw")
tmp <- (deviation(plotvalues()$all_vals))$DEVIATION
else if (input$devtrans == "absolute")
tmp <- abs((deviation(plotvalues()$all_vals))$DEVIATION)
else
tmp <- (deviation(plotvalues()$all_vals))$DEVIATION^2
mn <- signif(min(tmp[is.finite(tmp)]), 3)
mx <- signif(max(tmp[is.finite(tmp)]), 3)
sliderInput(inputId = "xrange_deviation", label = "x-axis limits",
min = mn, max = mx, value = c(mn, mx))
}
})
output$plot.deviation <- renderUI({
plotOutput("deviation", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "deviation_plot_click")
})
## Figure for exporting
output$export.deviation <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
print(plot_deviation(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_deviation,
plottype = input$devtype,
dojitter = input$dojitter,
transf = input$devtrans))
grDevices::dev.off()
})
## Data for exporting
output$export.deviation.df.rdata <- downloadHandler(
filename = "deviation-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.deviation.df.tsv <- downloadHandler(
filename = "deviation-data.tsv",
content = function(file) {
deviation_df <- isolate(deviation(plotvalues()$all_vals))
utils::write.table(deviation_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$deviation <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(deviation(plotvalues()$all_vals)),
paste0("Deviation plots can not be displayed. Check that ",
"cont_truth is not 'none' and that scores ",
"are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
plot_deviation(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_deviation,
plottype = input$devtype, dojitter = input$dojitter,
transf = input$devtrans)
})
})
output$deviation_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(deviation(plotvalues()$all_vals))
if (input$devtrans == "absolute")
all_data$absDEVIATION <- abs(all_data$DEVIATION)
if (input$devtrans == "squared")
all_data$sqDEVIATION <- all_data$DEVIATION^2
tmp_click <- input$deviation_plot_click
if (!is.null(tmp_click))
tmp_click$y <- round(tmp_click$y)
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, tmp_click,
threshold = 50, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method",
xvar = ifelse(input$devtrans == "raw", "DEVIATION",
ifelse(input$devtrans == "absolute",
"absDEVIATION", "sqDEVIATION")))
} else {
res <- nearPoints(all_data, tmp_click,
threshold = 50, maxpoints = 100,
addDist = TRUE, yvar = "num_method",
xvar = ifelse(input$devtrans == "raw", "DEVIATION",
ifelse(input$devtrans == "absolute",
"absDEVIATION", "sqDEVIATION")))
}
fix_res(res, methodcol = "fullmethod",
aspcts = ifelse(input$devtrans == "raw", "DEVIATION",
ifelse(input$devtrans == "absolute",
"absDEVIATION", "sqDEVIATION")),
tabtype = "deviation")
})
## ---------------------------- FPR -------------------------------- ##
output$plot.fpr <- renderUI({
plotOutput("fpr", width = "100%", height = paste0(input$plotheight, "px"),
hover = "fpr_plot_click")
})
## Figure for exporting
output$export.fpr <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpr(plotvalues()$all_vals)))
return(NULL)
print(plot_fpr(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fpr))
grDevices::dev.off()
})
## Data for exporting
output$export.fpr.df.rdata <- downloadHandler(
filename = "fpr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.fpr.df.tsv <- downloadHandler(
filename = "fpr-data.tsv",
content = function(file) {
fpr_df <- isolate(fpr(plotvalues()$all_vals))
utils::write.table(fpr_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fpr <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(fpr(plotvalues()$all_vals)),
paste0("FPR can not be displayed. Check that ",
"binary_truth is not 'none' and that adjusted ",
"p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpr(plotvalues()$all_vals)))
return(NULL)
plot_fpr(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, xaxisrange = input$xrange_fpr)
})
})
output$fpr_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpr(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(fpr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$fpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method", xvar = "FPR")
} else {
res <- nearPoints(all_data, input$fpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, yvar = "num_method", xvar = "FPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("FPR"),
tabtype = "large")
})
## ---------------------------- ROC -------------------------------- ##
output$plot.roc <- renderUI({
plotOutput("roc", width = "100%", height = paste0(input$plotheight, "px"),
hover = "roc_plot_click")
})
## Figure for exporting
output$export.roc <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(roc(plotvalues()$all_vals)))
return(NULL)
print(plot_roc(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_roc,
yaxisrange = input$yrange_roc,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.roc.df.rdata <- downloadHandler(
filename = "roc-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.roc.df.tsv <- downloadHandler(
filename = "roc-data.tsv",
content = function(file) {
roc_df <- isolate(roc(plotvalues()$all_vals))
utils::write.table(roc_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$roc <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(roc(plotvalues()$all_vals)),
paste0("ROC curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(roc(plotvalues()$all_vals)))
return(NULL)
plot_roc(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_roc, yaxisrange = input$yrange_roc,
linewidth = 1)
})
})
output$roc_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(roc(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(roc(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$roc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FPR", yvar = "TPR")
} else {
res <- nearPoints(all_data, input$roc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FPR", yvar = "TPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("FPR", "TPR"),
tabtype = "small")
})
## -------------------------- SCATTER ------------------------------ ##
output$plot.scatter <- renderUI({
plotOutput("scatter", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "scatter_plot_click")
})
## Figure for exporting
output$export.scatter <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(scatter(plotvalues()$all_vals)))
return(NULL)
print(plot_scatter(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, doflip = input$doflip,
dolog = input$dolog))
grDevices::dev.off()
})
## Data for exporting
output$export.scatter.df.rdata <- downloadHandler(
filename = "scatter-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.scatter.df.tsv <- downloadHandler(
filename = "scatter-data.tsv",
content = function(file) {
scatter_df <- isolate(scatter(plotvalues()$all_vals))
utils::write.table(scatter_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$scatter <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(scatter(plotvalues()$all_vals)),
paste0("Scatter plots can not be displayed. Check that ",
"cont_truth is not 'none' and that scores ",
"are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(scatter(plotvalues()$all_vals)))
return(NULL)
plot_scatter(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, doflip = input$doflip,
dolog = input$dolog)
})
})
output$scatter_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(scatter(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(scatter(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$scatter_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "OBSERVATION",
yvar = "TRUTH")
} else {
res <- nearPoints(all_data, input$scatter_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "OBSERVATION", yvar = "TRUTH")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("observation", "truth"),
tabtype = "scatter")
})
## ---------------------------- FPC -------------------------------- ##
output$plot.fpc <- renderUI({
plotOutput("fpc", width = "100%", height = paste0(input$plotheight, "px"),
hover = "fpc_plot_click")
})
## Figure for exporting
output$export.fpc <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpc(plotvalues()$all_vals)))
return(NULL)
print(plot_fpc(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
maxnfdc = input$maxnfdc,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fpc.df.rdata <- downloadHandler(
filename = "fpc-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.fpc.df.tsv <- downloadHandler(
filename = "fpc-data.tsv",
content = function(file) {
fpc_df <- isolate(fpc(plotvalues()$all_vals))
utils::write.table(fpc_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fpc <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(fpc(plotvalues()$all_vals)),
paste0("False discovery curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpc(plotvalues()$all_vals)))
return(NULL)
plot_fpc(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
maxnfdc = input$maxnfdc,
linewidth = 1)
})
})
output$fpc_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpc(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(fpc(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$fpc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "topN", yvar = "FP")
} else {
res <- nearPoints(all_data, input$fpc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "topN", yvar = "FP")
}
fix_res(res, methodcol = "fullmethod",
aspcts = c("Number of detections", "FP"), tabtype = "small")
})
## -------------------------- FDRTPR ------------------------------- ##
output$plot.fdrtprcurve <- renderUI({
plotOutput("fdrtprcurve", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "fdrtprcurve_plot_click")
})
## Figure for exporting
output$export.fdrtprcurve <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fdrtprcurve(plotvalues()$all_vals)))
return(NULL)
print(plot_fdrtprcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrtpr,
yaxisrange = input$yrange_fdrtpr,
plottype = input$plottype,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fdrtpr.df.rdata <- downloadHandler(
filename = "fdrtpr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$fdrtpr.df.tsv.button <- renderUI({
if ("curve" %in% input$plottype)
downloadButton("export.fdrtpr.df.tsv", label = "Download tsv (curve)")
else
downloadButton("export.fdrtpr.df.tsv", label = "Download tsv (points)")
})
output$export.fdrtpr.df.tsv <- downloadHandler(
filename = "fdrtpr-data.tsv",
content = function(file) {
fdrtpr_curve_df <- isolate(fdrtprcurve(plotvalues()$all_vals))
fdrtpr_point_df <- isolate(fdrtpr(plotvalues()$all_vals))
if ("curve" %in% input$plottype)
utils::write.table(fdrtpr_curve_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
else if ("points" %in% input$plottype)
utils::write.table(fdrtpr_point_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fdrtprcurve <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(any(c("points", "curve") %in% input$plottype),
"No plot type selected.")
)
if ("curve" %in% input$plottype)
validate(
need(is_plottable(fdrtprcurve(plotvalues()$all_vals)),
paste0("FDR/TPR curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
if ("points" %in% input$plottype)
validate(
need(is_plottable(fdrtpr(plotvalues()$all_vals)),
paste0("FDR/TPR at specified adjusted p-value thresholds ",
"can not be displayed. Check that binary_truth is not ",
"'none' and that adjusted p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrtprcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrtpr(plotvalues()$all_vals)))
return(NULL)
if (!any(c("curve", "points") %in% input$plottype))
return(NULL)
plot_fdrtprcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrtpr,
yaxisrange = input$yrange_fdrtpr,
plottype = input$plottype,
linewidth = 1)
})
})
output$fdrtprcurve_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrtprcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrtpr(plotvalues()$all_vals)))
return(NULL)
if ("curve" %in% input$plottype)
tab_data <- isolate(fdrtprcurve(plotvalues()$all_vals))
else tab_data <- isolate(fdrtpr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(tab_data, input$fdrtprcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FDR", yvar = "TPR")
} else {
res <- nearPoints(tab_data, input$fdrtprcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FDR", yvar = "TPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("FDR", "TPR"),
tabtype = "large")
})
## -------------------------- FDRNBR ------------------------------- ##
output$plot.fdrnbrcurve <- renderUI({
plotOutput("fdrnbrcurve", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "fdrnbrcurve_plot_click")
})
## Figure for exporting
output$export.fdrnbrcurve <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fdrnbrcurve(plotvalues()$all_vals)))
return(NULL)
print(plot_fdrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrnbr,
plottype = input$plottype,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fdrnbr.df.rdata <- downloadHandler(
filename = "fdrnbr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$fdrnbr.df.tsv.button <- renderUI({
if ("curve" %in% input$plottype)
downloadButton("export.fdrnbr.df.tsv", label = "Download tsv (curve)")
else
downloadButton("export.fdrnbr.df.tsv", label = "Download tsv (points)")
})
output$export.fdrnbr.df.tsv <- downloadHandler(
filename = "fdrnbr-data.tsv",
content = function(file) {
fdrnbr_curve_df <- isolate(fdrnbrcurve(plotvalues()$all_vals))
fdrnbr_point_df <- isolate(fdrnbr(plotvalues()$all_vals))
if ("curve" %in% input$plottype)
utils::write.table(fdrnbr_curve_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
else if ("points" %in% input$plottype)
utils::write.table(fdrnbr_point_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fdrnbrcurve <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(any(c("points", "curve") %in% input$plottype),
"No plot type selected.")
)
if ("curve" %in% input$plottype)
validate(
need(is_plottable(fdrnbrcurve(plotvalues()$all_vals)),
paste0("FDR/NBR curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
if ("points" %in% input$plottype)
validate(
need(is_plottable(fdrnbr(plotvalues()$all_vals)),
paste0("FDR/NBR at specified adjusted p-value thresholds ",
"can not be displayed. Check that binary_truth is not ",
"'none' and that adjusted p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrnbr(plotvalues()$all_vals)))
return(NULL)
if (!any(c("curve", "points") %in% input$plottype))
return(NULL)
plot_fdrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrnbr,
plottype = input$plottype,
linewidth = 1)
})
})
output$fdrnbrcurve_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrnbr(plotvalues()$all_vals)))
return(NULL)
if ("curve" %in% input$plottype)
tab_data <- isolate(fdrnbrcurve(plotvalues()$all_vals))
else tab_data <- isolate(fdrnbr(plotvalues()$all_vals))
if ("split" %in% input$facet_opt) {
res <- nearPoints(tab_data, input$fdrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FDR", yvar = "NBR")
} else {
res <- nearPoints(tab_data, input$fdrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FDR", yvar = "NBR")
}
methodcol <- "fullmethod"
fix_res(res, methodcol = methodcol,
aspcts = c("FDR", "Number of detections"), tabtype = "large")
})
## -------------------------- FSRNBR ------------------------------- ##
output$plot.fsrnbrcurve <- renderUI({
plotOutput("fsrnbrcurve", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "fsrnbrcurve_plot_click")
})
## Figure for exporting
output$export.fsrnbrcurve <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fsrnbrcurve(plotvalues()$all_vals)))
return(NULL)
print(plot_fsrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fsrnbr,
plottype = input$plottype,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fsrnbr.df.rdata <- downloadHandler(
filename = "fsrnbr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$fsrnbr.df.tsv.button <- renderUI({
if ("curve" %in% input$plottype)
downloadButton("export.fsrnbr.df.tsv", label = "Download tsv (curve)")
else
downloadButton("export.fsrnbr.df.tsv", label = "Download tsv (points)")
})
output$export.fsrnbr.df.tsv <- downloadHandler(
filename = "fsrnbr-data.tsv",
content = function(file) {
fsrnbr_curve_df <- isolate(fsrnbrcurve(plotvalues()$all_vals))
fsrnbr_point_df <- isolate(fsrnbr(plotvalues()$all_vals))
if ("curve" %in% input$plottype)
utils::write.table(fsrnbr_curve_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
else if ("points" %in% input$plottype)
utils::write.table(fsrnbr_point_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fsrnbrcurve <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(any(c("points", "curve") %in% input$plottype),
"No plot type selected.")
)
if ("curve" %in% input$plottype)
validate(
need(is_plottable(fsrnbrcurve(plotvalues()$all_vals)),
paste0("FSR/NBR curves can not be displayed. Check that ",
"cont_truth is not 'none' and that s-values ",
"are provided."))
)
if ("points" %in% input$plottype)
validate(
need(is_plottable(fsrnbr(plotvalues()$all_vals)),
paste0("FSR/NBR at specified adjusted p-value thresholds ",
"can not be displayed. Check that cont_truth is not ",
"'none' and that s-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fsrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fsrnbr(plotvalues()$all_vals)))
return(NULL)
if (!any(c("curve", "points") %in% input$plottype))
return(NULL)
plot_fsrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fsrnbr,
plottype = input$plottype,
linewidth = 1)
})
})
output$fsrnbrcurve_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fsrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fsrnbr(plotvalues()$all_vals)))
return(NULL)
if ("curve" %in% input$plottype)
tab_data <- isolate(fsrnbrcurve(plotvalues()$all_vals))
else tab_data <- isolate(fsrnbr(plotvalues()$all_vals))
if ("split" %in% input$facet_opt) {
res <- nearPoints(tab_data, input$fsrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FSR", yvar = "NBR")
} else {
res <- nearPoints(tab_data, input$fsrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FSR", yvar = "NBR")
}
methodcol <- "fullmethod"
fix_res(res, methodcol = methodcol,
aspcts = c("FSR", "Number of detections"), tabtype = "fsrnbr")
})
## Close app
shiny::observeEvent(input$close_app, {
shiny::stopApp()
})
} ## end of server_function
#nocov end
shinyApp(ui = p_layout, server = server_function)
}
markrobinsonuzh/iCOBRA documentation built on March 28, 2024, 2:01 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions COBRAapp
Documented in COBRAapp
#' Interactive shiny app to visualize results
#'
#' Interactive shiny app for visualization of results. The app can be
#' initialized with a \code{COBRAData} object. If no object is provided, truth
#' and results are loaded into the app from text files (see the Instructions tab
#' of the app for formatting instructions). Properly formatted text files can
#' also be obtained using the function \code{\link{COBRAData_to_text}}.
#'
#' @param cobradata An (optional) \code{COBRAData} object. If not given, the
#' user can load results from text files.
#' @param autorun A logical indicating whether the app calculations should start
#' automatically on launch, or wait for the user to press the 'Start
#' calculation!' button.
#' @author Charlotte Soneson
#' @return Returns (and runs) an object representing the shiny app.
#' @import shiny
#' @import shinydashboard
#' @importFrom shinyBS bsTooltip
#' @importFrom utils packageVersion
#' @export
#' @examples
#' data(cobradata_example)
#' app <- COBRAapp(cobradata_example)
#' if (interactive()) {
#' shiny::runApp(app)
#' }
COBRAapp <- function(cobradata = NULL, autorun = FALSE) {
## ------------------------------------------------------------------ ##
## Define UI ##
## ------------------------------------------------------------------ ##
p_layout <-
shinydashboard::dashboardPage(
skin = "blue",
shinydashboard::dashboardHeader(
title = paste0("iCOBRA - interactive COmparative evaluation of ",
"Binary classification and RAnking methods (v",
packageVersion("iCOBRA"), ")"),
titleWidth = 950),
shinydashboard::dashboardSidebar(
width = 350,
## Settings and inputs for the truth
shinydashboard::menuItem("Truth", icon = icon("database"),
## Load the file containing the truth.
uiOutput("choose_truth_file"),
shinyBS::bsTooltip(
"truth", paste0("Select the file containing the true status",
" of each feature. See the Instructions ",
"tab for formatting instructions."),
"right", options = list(container = "body")),
## Define the column containing the feature identifier
## (for both truth and results)
uiOutput("choose_feature_id"),
## Define the column containing the (binary) truth
uiOutput("choose_binary_truth"),
## Define the column containing the (continuous) truth
uiOutput("choose_continuous_truth"),
## Define the variable used to stratify the results,
## based on the columns available in the truth file.
uiOutput("splitvar"),
shinyBS::bsTooltip(
"splitvar",
paste0("Select a feature attribute by which to stratify ",
"the result representations."),
"right", options = list(container = "body")),
## Define the maximal number of categories to retain
## in stratification.
uiOutput("choosemaxsplit"),
shinyBS::bsTooltip(
"choosemaxsplit",
paste0("Set the number of categories to show if the results",
" are stratified by a variable annotation. The most ",
"frequent categories with both positive and negative",
" instances will be retained"),
"right", options = list(container = "body")),
## Decide whether or not to include the "overall" category when
## stratifying the results.
uiOutput("chooseincludeoverall"),
shinyBS::bsTooltip(
"includeoverall",
paste0("Select whether or not to include the 'overall' ",
"class when showing stratified results."),
"right", options = list(container = "body"))),
## Settings and inputs for results
shinydashboard::menuItem("Results", icon = icon("folder-open"),
## Load the file containing the results.
uiOutput("choose_result_file"),
shinyBS::bsTooltip(
"file1", paste0("Select file containing results from one or",
" multiple methods. See the Instructions ",
"tab for formatting instructions."),
"right", options = list(container = "body")),
## Decide which methods to include in the results.
## Depends on the loaded result files.
uiOutput("columns"),
shinyBS::bsTooltip(
"columns",
paste0("Select the methods for which to show the results."),
"right", options = list(container = "body")),
## Decide whether to base evaluations on all genes in
## truth table, or only on variables for which both a truth
## value and a result is provided
checkboxInput(
inputId = "onlyshared",
label = paste0("Calculate performance based only on ",
"features shared between truth and result ",
"tables."),
value = FALSE),
shinyBS::bsTooltip(
"onlyshared",
paste0("Calculate performance based only on ",
"features shared between truth and result ",
"tables. Otherwise, all features in the ",
"truth table will be used."),
"right", options = list(container = "body")),
actionButton("goButton", "Start calculation!",
icon = icon("plane"))),
## Plot settings
shinydashboard::menuItem("Plot settings", icon = icon("paint-brush"),
## Choose color scheme
selectInput(
"colorscheme", "Select color palette",
c("hue_pal", "Accent (max 7 methods)",
"Dark2 (max 7 methods)", "Paired (max 11 methods)",
"Pastel1 (max 8 methods)", "Pastel2 (max 7 methods)",
"Set1 (max 8 methods)", "Set2 (max 7 methods)",
"Set3 (max 11 methods)", "rainbow", "heat", "terrain",
"topo", "cm"),
selectize = TRUE),
shinyBS::bsTooltip(
"colorscheme",
paste0("Choose color palette. Some palettes are only ",
"applicable if the number of methods ",
"(or method/stratification level combinations) is ",
"below a certain threshold. If this threshold is ",
"exceeded, the colorscheme will default to hue_pal."),
"right", options = list(container = "body")),
## Decide what to include in FDR/TPR plots.
checkboxGroupInput(
"plottype",
paste0("Display full curve and/or points in FDR/TPR, ",
"FDR/NBR and FSR/NBR plots"),
c("curve", "points"), c("points")),
## Decide whether or not to facet plots.
checkboxGroupInput(
"facet_opt",
"Split plots into panels by stratifying variable",
"split", "split"),
## Define the q-value thresholds to use in the plots.
textInput(inputId = "fdrthresholds", label = "FDR thresholds",
value = "0.01, 0.05, 0.1"),
shinyBS::bsTooltip("fdrthresholds",
paste0("Specific FDR thresholds at which the ",
"performance will be evaluated. ",
"Separate multiple values with comma"),
"right", options = list(container = "body")),
## Define the s-value thresholds to use in the plots.
textInput(inputId = "svalthresholds",
label = "s-value thresholds",
value = "0.01, 0.05, 0.1"),
shinyBS::bsTooltip("svalthresholds",
paste0("Specific s-value thresholds at ",
"which the performance will be ",
"evaluated. Separate multiple ",
"values with comma"),
"right",
options = list(container = "body")),
## Define the plot height (in pixels).
numericInput(inputId = "plotheight",
label = "Plot height (numeric, in pixels)",
value = 600, min = 200, max = 2000, step = 10),
shinyBS::bsTooltip("plotheight",
paste0("The height of the plots (in pixels). ",
"Default 800."),
"right", options = list(container = "body")),
## Define the pointsize used in the plots.
numericInput(inputId = "pointsize", label = "Point size",
value = 5),
shinyBS::bsTooltip("pointsize",
paste0("The point size used in the plots."),
"right", options = list(container = "body")),
## Define the fontsize used in the panel headers.
numericInput(inputId = "stripsize",
label = "Font size for panel headers",
value = 15),
shinyBS::bsTooltip("stripsize",
paste0("The font size used for panel headers in",
" facetted plots."),
"right", options = list(container = "body"))),
## Button to close app
shiny::actionButton("close_app", "Close app")
),
## Outputs
shinydashboard::dashboardBody(fluidRow(
shinydashboard::tabBox(
width = 12,
## Define output size of error messages
tags$head(
tags$style(HTML("
.shiny-output-error-validation {
font-size: 17px;
}
"))
),
tabPanel("About",
includeMarkdown(system.file("extdata", "about.md",
package = "iCOBRA")),
value = "about"),
tabPanel("Instructions",
includeMarkdown(system.file("extdata", "instructions.md",
package = "iCOBRA")),
value = "instructions"),
tabPanel("Data preview",
h1("Truth"),
DT::dataTableOutput("table.truth"),
h1("p-values"),
DT::dataTableOutput("table.pval"),
h1("adjusted p-values"),
DT::dataTableOutput("table.padj"),
h1("s-values"),
DT::dataTableOutput("table.sval"),
h1("scores"),
DT::dataTableOutput("table.score"),
value = "datapreview"),
tabPanel("TPR vs FDR",
uiOutput("plot.fdrtprcurve"),
fluidRow(
column(3, sliderInput(inputId = "xrange_fdrtpr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(3, sliderInput(inputId = "yrange_fdrtpr",
label = "y-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fdrtprcurve",
label = "Download plot")),
column(2, br(), downloadButton("export.fdrtpr.df.rdata",
label = "Download Rdata")),
column(2, br(), uiOutput("fdrtpr.df.tsv.button"))
),
DT::dataTableOutput("fdrtprcurve_click_info"),
value = "fdrtprcurve"),
tabPanel("NBR vs FDR",
uiOutput("plot.fdrnbrcurve"),
fluidRow(
column(4, sliderInput(inputId = "xrange_fdrnbr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fdrnbrcurve",
label = "Download plot")),
column(2, br(), downloadButton("export.fdrnbr.df.rdata",
label = "Download Rdata")),
column(2, br(), uiOutput("fdrnbr.df.tsv.button"))
),
DT::dataTableOutput("fdrnbrcurve_click_info"),
value = "fdrnbrcurve"),
tabPanel("TPR", uiOutput("plot.tpr"),
fluidRow(
column(4, sliderInput(inputId = "xrange_tpr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.tpr",
label = "Download plot")),
column(2, br(), downloadButton("export.tpr.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.tpr.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("tpr_click_info"), value = "tpr"),
tabPanel("FPR", uiOutput("plot.fpr"),
fluidRow(
column(4, sliderInput(inputId = "xrange_fpr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fpr",
label = "Download plot")),
column(2, br(), downloadButton("export.fpr.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.fpr.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("fpr_click_info"), value = "fpr"),
tabPanel("ROC", uiOutput("plot.roc"),
fluidRow(
column(3, sliderInput(inputId = "xrange_roc",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(3, sliderInput(inputId = "yrange_roc",
label = "y-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.roc",
label = "Download plot")),
column(2, br(), downloadButton("export.roc.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.roc.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("roc_click_info"), value = "roc"),
tabPanel("False discovery curves",
uiOutput("plot.fpc"),
fluidRow(
column(3, numericInput(inputId = "maxnfdc",
label = "Maximal rank to display",
value = 500, min = 10, max = 10000,
step = 10)),
column(2, br(), downloadButton("export.fpc",
label = "Download plot")),
column(2, br(), downloadButton("export.fpc.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.fpc.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("fpc_click_info"), value = "fpc"),
tabPanel("Correlation",
uiOutput("plot.corr"),
fluidRow(
column(3, sliderInput(inputId = "xrange_corr",
label = "x-axis limits",
min = -1, max = 1, value = c(-1, 1),
step = 0.01)),
column(2, radioButtons(inputId = "corrtype",
label = "Correlation measure",
choices = c("pearson", "spearman"),
selected = "spearman")),
column(2, br(), downloadButton("export.corr",
label = "Download plot")),
column(2, br(), downloadButton("export.corr.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.corr.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("corr_click_info"), value = "corr"),
tabPanel("Scatter",
uiOutput("plot.scatter"),
fluidRow(
column(2, checkboxInput("doflip", "Flip axes", FALSE),
checkboxInput("dolog", "Log-transform", FALSE)),
column(2, br(), downloadButton("export.scatter",
label = "Download plot")),
column(2, br(), downloadButton("export.scatter.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.scatter.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("scatter_click_info"),
value = "scatter"),
tabPanel("Deviations",
uiOutput("plot.deviation"),
fluidRow(
column(2, checkboxInput("dojitter",
"Include jittered points", FALSE),
radioButtons(inputId = "devtrans",
label = "Transformation",
choices = c("raw", "absolute",
"squared"),
selected = "raw")),
column(1, radioButtons(
inputId = "devtype", label = "Plot type",
choices = c("boxplot", "violinplot"),
selected = "violinplot")),
column(3, uiOutput("axislimitsdeviation")),
column(2, br(), downloadButton("export.deviation",
label = "Download plot")),
column(2, br(), downloadButton("export.deviation.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.deviation.df.tsv",
label = "Download tsv"))
),
DT::dataTableOutput("deviation_click_info"),
value = "deviation"),
tabPanel("Venn diagram",
uiOutput("plot.overlap"),
fluidRow(
column(1, radioButtons(inputId = "incltruth",
label = "Include truth",
choices = c("yes", "no"),
selected = "yes")),
shinyBS::bsTooltip(
"incltruth",
paste0("Whether or nor to include the truth as a ",
"(perfect) method in the Venn diagrams. Note ",
"that maximally five methods (including the ",
"truth) can be included."),
"bottom", options = list(container = "body")),
column(1, radioButtons(inputId = "vennType",
label = "Type of threshold",
choices = c("adjp", "rank"),
selected = "adjp")),
shinyBS::bsTooltip(
"vennType",
paste0("How to determine the set of features to ",
"compare between methods. "),
"bottom", options = list(container = "body")),
column(2, numericInput(
inputId = "adjpVenn",
label = "Adjusted p-value threshold",
value = 0.05, min = 0, max = 1, step = 0.01)),
shinyBS::bsTooltip(
"adjpVenn",
paste0("The adjusted p-value threshold used to extract ",
"the sets of significant variables to use for ",
"the Venn diagram. "),
"bottom", options = list(container = "body")),
column(2, numericInput(
inputId = "rankVenn",
label = "Rank threshold",
value = 100, min = 0, max = 1e10, step = 1)),
shinyBS::bsTooltip(
"rankVenn",
paste0("The rank used to extract the top-ranked ",
"features to compare in the Venn diagram. "),
"bottom", options = list(container = "body")),
column(2, br(), downloadButton("export.overlap",
label = "Download plot")),
column(2, br(), downloadButton("export.overlap.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.overlap.df.tsv",
label = "Download tsv"))
),
value = "overlap"),
tabPanel("UpSet plot",
uiOutput("plot.upset"),
fluidRow(
column(2, uiOutput("upset.stratum")),
column(2, radioButtons(inputId = "order.upset.by",
label = "Order intersections by",
choices = c("degree", "freq"),
selected = "freq")),
column(2, radioButtons(inputId = "upset.order",
label = "Order intersections",
choices = c("decreasing",
"increasing"),
selected = "decreasing")),
column(2, br(), downloadButton("export.upset",
label = "Download plot")),
column(2, br(), downloadButton("export.upset.df.rdata",
label = "Download Rdata")),
column(2, br(), downloadButton("export.upset.df.tsv",
label = "Download tsv"))
),
value = "upset"),
tabPanel("NBR vs FSR",
uiOutput("plot.fsrnbrcurve"),
fluidRow(
column(4, sliderInput(inputId = "xrange_fsrnbr",
label = "x-axis limits",
min = 0, max = 1, value = c(0, 1),
step = 0.01)),
column(2, br(), downloadButton("export.fsrnbrcurve",
label = "Download plot")),
column(2, br(), downloadButton("export.fsrnbr.df.rdata",
label = "Download Rdata")),
column(2, br(), uiOutput("fsrnbr.df.tsv.button"))
),
DT::dataTableOutput("fsrnbrcurve_click_info"),
value = "fsrnbrcurve"),
selected = "fdrtprcurve"
)
))
)
options(shiny.maxRequestSize = 15*1024^2)
## ------------------------------------------------------------------ ##
## Define server ##
## ------------------------------------------------------------------ ##
#nocov start
server_function <- function(input, output, session) {
values <- reactiveValues()
values$my_cobradata <- COBRAData()
output$choose_truth_file <- renderUI({
if (!is.null(cobradata)) {
NULL
} else {
return(fileInput(inputId = "truth", label = "Load truth file",
accept = c("text/csv", "text/comma-separated-values",
"text/tab-separated-values", "text/plain",
".csv", ".tsv", ".tab", ".txt"),
multiple = FALSE))
}
})
truthFile <- reactive({
if (is.null(input$truth))
return(NULL)
trf <- utils::read.delim(input$truth$datapath, header = TRUE,
as.is = TRUE, sep = "\t", quote = "",
check.names = FALSE)
isolate(values$my_cobradata <-
COBRAData(truth = trf,
object_to_extend = values$my_cobradata))
return(trf)
})
## Render the UI element to choose the column containing the feature ID
output$choose_feature_id <- renderUI({
if (is.null(truthFile()))
return(NULL)
selectInput("feature_id", "Select column containing feature identifiers",
colnames(truthFile()), selectize = TRUE)
})
## Render the UI element to choose the binary truth column
output$choose_binary_truth <- renderUI({
if (!is.null(cobradata)) {
ccl <- sapply(truth(cobradata), class)
only01 <- sapply(truth(cobradata), function(i) {
if (all(i %in% c(0, 1, NA))) "binary"
else "notbinary"
})
} else {
if (is.null(truthFile()))
return(NULL)
ccl <- sapply(truthFile(), class)
only01 <- sapply(truthFile(), function(i) {
if (all(i %in% c(0, 1, NA))) "binary"
else "notbinary"
})
}
keepcols <- intersect(which(ccl %in% c("numeric", "integer")),
which(only01 == "binary"))
if (!is.null(cobradata)) {
selectInput("binary_truth", "Select column containing binary truth",
c(colnames(truth(cobradata))[keepcols], "none"),
selectize = TRUE)
} else {
selectInput("binary_truth", "Select column containing binary truth",
c(colnames(truthFile())[keepcols], "none"),
selectize = TRUE)
}
})
## Render the UI element to choose the continuous truth column
output$choose_continuous_truth <- renderUI({
if (!is.null(cobradata)) {
ccl <- sapply(truth(cobradata), class)
} else {
if (is.null(truthFile()))
return(NULL)
ccl <- sapply(truthFile(), class)
}
keepcols <- which(ccl %in% c("numeric", "integer"))
if (!is.null(cobradata)) {
selectInput("cont_truth", "Select column containing continuous truth",
c("none", colnames(truth(cobradata))[keepcols]),
selectize = TRUE)
} else {
selectInput("cont_truth", "Select column containing continuous truth",
c("none", colnames(truthFile())[keepcols]),
selectize = TRUE)
}
})
## Render the UI element to choose which attribute (from the truth file)
## to split panels by.
output$splitvar <- renderUI({
if (!is.null(cobradata)) {
selectInput("splv", "Select variable to stratify by",
c("none", colnames(truth(cobradata))), selectize = TRUE)
} else {
trf <- truthFile()
if (is.null(trf))
return(NULL)
selectInput("splv", "Select variable to stratify by",
c("none", colnames(trf)), selectize = TRUE)
}
})
## Render the UI element to choose how many categories to include.
output$choosemaxsplit <- renderUI({
numericInput(inputId = "maxsplit",
label = "Maximum number of levels to show when stratifying",
value = 3, min = 1)
})
## Render the UI element to determine whether the overall category should
## be included when stratifying
output$chooseincludeoverall <- renderUI({
radioButtons(inputId = "includeoverall",
label = "Include 'overall' class when stratifying",
choices = c("yes", "no"), selected = "yes")
})
## Render the UI element to upload result file(s)
output$choose_result_file <- renderUI({
if (!is.null(cobradata)) {
NULL
} else {
return(fileInput(inputId = "file1", label = "Add file with results",
accept = c("text/csv", "text/comma-separated-values",
"text/tab-separated-values", "text/plain",
".csv", ".tsv"), multiple = FALSE))
}
})
## Render the UI object for selecting methods to include, based on the
## uploaded result files.
output$columns <- renderUI({
if (!is.null(cobradata)) {
allm <- unique(c(colnames(pval(cobradata)), colnames(padj(cobradata)),
colnames(sval(cobradata)), colnames(score(cobradata))))
isolate(values$my_cobradata <- cobradata)
isolate(values$all_methods <- allm)
isolate(values$my_cobradata <- calculate_adjp(values$my_cobradata))
## Create the UI element for selecting methods to include.
checkboxGroupInput("cols", "Select methods",
isolate(values$all_methods),
isolate(values$all_methods))
} else {
rownames(truth(values$my_cobradata)) <-
truth(values$my_cobradata)[, input$feature_id]
## Read file
inFile <- input$file1
if (is.null(inFile))
return(NULL)
v <- utils::read.delim(inFile$datapath, header = TRUE, as.is = TRUE,
sep = "\t", quote = "", check.names = FALSE)
## Perform some checks on the file, if ok then continue
if (res_check(v)) {
## Extract all columns corresponding to methods
ii <- 1
while(ii <= ncol(v)) {
i <- colnames(v)[ii]
coltype <- get_coltype(i)
if (!(is.null(coltype))) {
## Fix duplicates
tmp1 <- v[, c(input$feature_id, i)]
tmp1 <- fix_duplicates(tmp1, input$feature_id, i)
## Rename column to only method name
tmp <- tmp1[i]
rownames(tmp) <- tmp1[, input$feature_id]
i <- gsub(":S", "", gsub(":score$", "", gsub(":adjP$", "",
gsub(":P$", "", i))))
colnames(tmp) <- i
## Add values to data collection
isolate(values$all_methods <- unique(c(values$all_methods, i)))
if (coltype == "pval") {
isolate(values$my_cobradata <-
COBRAData(pval = tmp,
object_to_extend = values$my_cobradata))
} else if (coltype == "padj") {
isolate(values$my_cobradata <-
COBRAData(padj = tmp,
object_to_extend = values$my_cobradata))
} else if (coltype == "sval") {
isolate(values$my_cobradata <-
COBRAData(sval = tmp,
object_to_extend = values$my_cobradata))
} else if (coltype == "score") {
isolate(values$my_cobradata <-
COBRAData(score = tmp,
object_to_extend = values$my_cobradata))
}
}
ii <- ii + 1
}
## Calculate adjusted p-values if needed
isolate(values$my_cobradata <- calculate_adjp(values$my_cobradata))
## Render the UI element
checkboxGroupInput("cols", "Select methods",
isolate(values$all_methods),
isolate(values$all_methods))
}
}
})
plotvalues_TPR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
thrs <- sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$fdrthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "tpr",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FDR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
thrs <- sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$fdrthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fdrtpr",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FPR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
thrs <- sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$fdrthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fpr",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_ROC <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "roc",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FPC <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fpc",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FDRTPR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$binary_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = input$binary_truth,
cont_truth = NULL,
aspects = "fdrtprcurve",
thrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FSR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
thrs <-
sort(unique(as.numeric(gsub(" ", "",
unlist(strsplit(input$svalthresholds,
","))))))
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "fsrnbr",
svalthrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_FSRNBR <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "fsrnbrcurve",
svalthrs = thrs,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_overlap <- reactive({
if ((input$goButton > 0 | isTRUE(autorun))) {
if (input$binary_truth == "none") {
bt <- NULL
} else {
bt <- input$binary_truth
}
return(calculate_performance(values$my_cobradata,
binary_truth = bt,
cont_truth = NULL,
aspects = "overlap",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = input$adjpVenn,
type_venn = input$vennType,
topn_venn = input$rankVenn))
} else {
return(COBRAPerformance())
}
})
plotvalues_corr <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "corr",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_scatter <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "scatter",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues_deviation <- reactive({
if ((input$goButton > 0 | isTRUE(autorun)) &
input$cont_truth != "none") {
return(calculate_performance(values$my_cobradata,
binary_truth = NULL,
cont_truth = input$cont_truth,
aspects = "deviation",
thrs = NULL,
splv = input$splv,
maxsplit = input$maxsplit,
onlyshared = input$onlyshared,
thr_venn = NULL,
type_venn = "adjp",
topn_venn = NULL))
} else {
return(COBRAPerformance())
}
})
plotvalues <- reactive({
if ((input$goButton > 0 | isTRUE(autorun))) {
## Put together results for chosen methods
if (length(input$cols) != 0) {
withProgress(message = "Calculating...", value = 0, {
all_vals <-
COBRAPerformance(tpr = tpr(plotvalues_TPR()),
fdrtpr = fdrtpr(plotvalues_FDR()),
fdrnbr = fdrtpr(plotvalues_FDR()),
fdrtprcurve = fdrtprcurve(plotvalues_FDRTPR()),
fdrnbrcurve = fdrnbrcurve(plotvalues_FDRTPR()),
fsrnbr = fsrnbr(plotvalues_FSR()),
fsrnbrcurve = fsrnbrcurve(plotvalues_FSRNBR()),
fpr = fpr(plotvalues_FPR()),
roc = roc(plotvalues_ROC()),
fpc = fpc(plotvalues_FPC()),
overlap = overlap(plotvalues_overlap()),
maxsplit = input$maxsplit,
splv = input$splv,
corr = corr(plotvalues_corr()),
scatter = scatter(plotvalues_scatter()),
deviation = deviation(plotvalues_deviation()))
all_vals <- prepare_data_for_plot(
cobraperf = all_vals,
keepmethods = input$cols,
incloverall = (input$includeoverall == "yes"),
colorscheme = input$colorscheme,
facetted = ("split" %in% input$facet_opt),
incltruth = (input$incltruth == "yes"))
## This is a bit of a hack, just to make the plots directly
## depend on the truth input. Otherwise, they will not be
## automatically updated if the truth file is changed
## (not until something else that directly affects the plots
## is changed as well, like the FDR thresholds or plot size etc).
## The title is rendered in white, so it is not visible.
title <- input$truth$name
list(all_vals = all_vals, title = title)
})
}
} else {
NULL
}
})
## -------------------------- OVERLAP ------------------------------ ##
## Generate Venn diagram plot
output$plot.overlap <- renderUI({
plotOutput("overlap", width = "100%",
height = paste0(input$plotheight, "px"))
})
## Figure for exporting
output$export.overlap <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, height = 12, width = 12)
if (length(values$all_methods) == 0 | length(input$cols) == 0)
return(NULL)
plot_overlap(plotvalues()$all_vals)
grDevices::dev.off()
})
## Data for exporting
output$export.overlap.df.rdata <- downloadHandler(
filename = "overlap-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.overlap.df.tsv <- downloadHandler(
filename = "overlap-data.tsv",
content = function(file) {
overlap_df <- overlap(plotvalues()$all_vals)
if (is(overlap_df, "list")) {
overlap_df <- lapply(overlap_df,
function(w) cbind(feature = rownames(w), w))
for (i in seq_len(length(overlap_df))) {
overlap_df[[i]]$featureclass <- names(overlap_df)[i]
}
overlap_df <- do.call(rbind, overlap_df)
}
else {
overlap_df <- cbind(feature = rownames(overlap_df), overlap_df)
}
utils::write.table(overlap_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$overlap <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(is_plottable(overlap(plotvalues()$all_vals)),
paste0("Venn diagrams can not be displayed. Check that ",
"binary_truth is not 'none', and that adjusted ",
"p-values are provided."))
)
validate(
need((is(overlap(plotvalues()$all_vals), "list") &&
ncol(overlap(plotvalues()$all_vals)[[1]]) <= 5) |
(is(overlap(plotvalues()$all_vals), "data.frame") &&
ncol(overlap(plotvalues()$all_vals)) <= 5),
paste0("Venn diagrams can not be constructed if more than five ",
"methods are selected (including the truth, ",
"if applicable)."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun))) {
return(NULL)
}
plot_overlap(plotvalues()$all_vals)
})
})
## --------------------------- UPSET ------------------------------- ##
## Choose category
output$upset.stratum <- renderUI({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(overlap(plotvalues()$all_vals)))
return(NULL)
else {
if (!is(overlap(plotvalues()$all_vals), "list"))
selectInput("upset.stratum", "Select category",
"overall", selectize = TRUE)
else
selectInput("upset.stratum", "Select category",
names(overlap(plotvalues()$all_vals)),
selectize = TRUE)
}
})
## Generate UpSet plot
output$plot.upset <- renderUI({
plotOutput("upset", width = "100%",
height = paste0(input$plotheight, "px"))
})
## Figure for exporting
output$export.upset <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, height = 12, width = 12)
if (length(values$all_methods) == 0 | length(input$cols) == 0)
return(NULL)
plot_upset(plotvalues()$all_vals, order.by = input$order.upset.by,
decreasing = (input$upset.order == "decreasing"),
stratum = input$upset.stratum)
grDevices::dev.off()
})
## Data for exporting
output$export.upset.df.rdata <- downloadHandler(
filename = "upset-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.upset.df.tsv <- downloadHandler(
filename = "upset-data.tsv",
content = function(file) {
overlap_df <- overlap(plotvalues()$all_vals)
if (is(overlap_df, "list")) {
overlap_df <- lapply(overlap_df,
function(w) cbind(feature = rownames(w), w))
for (i in seq_len(length(overlap_df))) {
overlap_df[[i]]$featureclass <- names(overlap_df)[i]
}
overlap_df <- do.call(rbind, overlap_df)
}
else {
overlap_df <- cbind(feature = rownames(overlap_df), overlap_df)
}
utils::write.table(overlap_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$upset <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(is_plottable(overlap(plotvalues()$all_vals)),
paste0("UpSet plots can not be displayed. Check that ",
"binary_truth is not 'none', and that adjusted ",
"p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun))) {
return(NULL)
}
plot_upset(plotvalues()$all_vals, order.by = input$order.upset.by,
decreasing = (input$upset.order == "decreasing"),
stratum = input$upset.stratum)
})
})
## ---------------------------- TPR -------------------------------- ##
output$plot.tpr <- renderUI({
plotOutput("tpr", width = "100%", height = paste0(input$plotheight, "px"),
hover = "tpr_plot_click")
})
## Figure for exporting
output$export.tpr <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(tpr(plotvalues()$all_vals)))
return(NULL)
print(plot_tpr(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_tpr))
grDevices::dev.off()
})
## Data for exporting
output$export.tpr.df.rdata <- downloadHandler(
filename = "tpr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.tpr.df.tsv <- downloadHandler(
filename = "tpr-data.tsv",
content = function(file) {
tpr_df <- isolate(tpr(plotvalues()$all_vals))
utils::write.table(tpr_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$tpr <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(tpr(plotvalues()$all_vals)),
paste0("TPR can not be displayed. Check that ",
"binary_truth is not 'none' and that adjusted ",
"p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(tpr(plotvalues()$all_vals))) {
return(NULL)
}
plot_tpr(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, xaxisrange = input$xrange_tpr)
})
})
output$tpr_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(tpr(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(tpr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$tpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method", xvar = "TPR")
} else {
res <- nearPoints(all_data, input$tpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, yvar = "num_method", xvar = "TPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("TPR"),
tabtype = "large")
})
## ------------------------ DATA PREVIEW --------------------------- ##
output$table.truth <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(truth(values$my_cobradata))
})
output$table.pval <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(pval(values$my_cobradata))
})
output$table.padj <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(padj(values$my_cobradata))
})
output$table.sval <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(sval(values$my_cobradata))
})
output$table.score <- DT::renderDataTable({
if (length(values$all_methods) == 0)
return(NULL)
DT::datatable(score(values$my_cobradata))
})
## ---------------------------- CORR ------------------------------- ##
output$plot.corr <- renderUI({
plotOutput("corr", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "corr_plot_click")
})
## Figure for exporting
output$export.corr <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(corr(plotvalues()$all_vals)))
return(NULL)
print(plot_corr(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_corr,
corrtype = input$corrtype))
grDevices::dev.off()
})
## Data for exporting
output$export.corr.df.rdata <- downloadHandler(
filename = "corr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.corr.df.tsv <- downloadHandler(
filename = "corr-data.tsv",
content = function(file) {
tpr_df <- isolate(corr(plotvalues()$all_vals))
utils::write.table(tpr_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$corr <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(corr(plotvalues()$all_vals)),
paste0("Correlation plots can not be displayed. Check that ",
"cont_truth is not 'none' and that scores ",
"are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(corr(plotvalues()$all_vals)))
return(NULL)
plot_corr(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, xaxisrange = input$xrange_corr,
corrtype = input$corrtype)
})
})
output$corr_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(corr(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(corr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$corr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method", xvar = toupper(input$corrtype))
} else {
res <- nearPoints(all_data, input$corr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, yvar = "num_method",
xvar = toupper(input$corrtype))
}
fix_res(res, methodcol = "fullmethod", aspcts = toupper(input$corrtype),
tabtype = "corr")
})
## -------------------------- DEVIATION ---------------------------- ##
output$axislimitsdeviation <- renderUI({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
else {
if (input$devtrans == "raw")
tmp <- (deviation(plotvalues()$all_vals))$DEVIATION
else if (input$devtrans == "absolute")
tmp <- abs((deviation(plotvalues()$all_vals))$DEVIATION)
else
tmp <- (deviation(plotvalues()$all_vals))$DEVIATION^2
mn <- signif(min(tmp[is.finite(tmp)]), 3)
mx <- signif(max(tmp[is.finite(tmp)]), 3)
sliderInput(inputId = "xrange_deviation", label = "x-axis limits",
min = mn, max = mx, value = c(mn, mx))
}
})
output$plot.deviation <- renderUI({
plotOutput("deviation", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "deviation_plot_click")
})
## Figure for exporting
output$export.deviation <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
print(plot_deviation(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_deviation,
plottype = input$devtype,
dojitter = input$dojitter,
transf = input$devtrans))
grDevices::dev.off()
})
## Data for exporting
output$export.deviation.df.rdata <- downloadHandler(
filename = "deviation-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.deviation.df.tsv <- downloadHandler(
filename = "deviation-data.tsv",
content = function(file) {
deviation_df <- isolate(deviation(plotvalues()$all_vals))
utils::write.table(deviation_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$deviation <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(deviation(plotvalues()$all_vals)),
paste0("Deviation plots can not be displayed. Check that ",
"cont_truth is not 'none' and that scores ",
"are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
plot_deviation(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_deviation,
plottype = input$devtype, dojitter = input$dojitter,
transf = input$devtrans)
})
})
output$deviation_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(deviation(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(deviation(plotvalues()$all_vals))
if (input$devtrans == "absolute")
all_data$absDEVIATION <- abs(all_data$DEVIATION)
if (input$devtrans == "squared")
all_data$sqDEVIATION <- all_data$DEVIATION^2
tmp_click <- input$deviation_plot_click
if (!is.null(tmp_click))
tmp_click$y <- round(tmp_click$y)
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, tmp_click,
threshold = 50, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method",
xvar = ifelse(input$devtrans == "raw", "DEVIATION",
ifelse(input$devtrans == "absolute",
"absDEVIATION", "sqDEVIATION")))
} else {
res <- nearPoints(all_data, tmp_click,
threshold = 50, maxpoints = 100,
addDist = TRUE, yvar = "num_method",
xvar = ifelse(input$devtrans == "raw", "DEVIATION",
ifelse(input$devtrans == "absolute",
"absDEVIATION", "sqDEVIATION")))
}
fix_res(res, methodcol = "fullmethod",
aspcts = ifelse(input$devtrans == "raw", "DEVIATION",
ifelse(input$devtrans == "absolute",
"absDEVIATION", "sqDEVIATION")),
tabtype = "deviation")
})
## ---------------------------- FPR -------------------------------- ##
output$plot.fpr <- renderUI({
plotOutput("fpr", width = "100%", height = paste0(input$plotheight, "px"),
hover = "fpr_plot_click")
})
## Figure for exporting
output$export.fpr <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpr(plotvalues()$all_vals)))
return(NULL)
print(plot_fpr(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fpr))
grDevices::dev.off()
})
## Data for exporting
output$export.fpr.df.rdata <- downloadHandler(
filename = "fpr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.fpr.df.tsv <- downloadHandler(
filename = "fpr-data.tsv",
content = function(file) {
fpr_df <- isolate(fpr(plotvalues()$all_vals))
utils::write.table(fpr_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fpr <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(fpr(plotvalues()$all_vals)),
paste0("FPR can not be displayed. Check that ",
"binary_truth is not 'none' and that adjusted ",
"p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpr(plotvalues()$all_vals)))
return(NULL)
plot_fpr(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, xaxisrange = input$xrange_fpr)
})
})
output$fpr_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpr(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(fpr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$fpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
yvar = "num_method", xvar = "FPR")
} else {
res <- nearPoints(all_data, input$fpr_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, yvar = "num_method", xvar = "FPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("FPR"),
tabtype = "large")
})
## ---------------------------- ROC -------------------------------- ##
output$plot.roc <- renderUI({
plotOutput("roc", width = "100%", height = paste0(input$plotheight, "px"),
hover = "roc_plot_click")
})
## Figure for exporting
output$export.roc <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(roc(plotvalues()$all_vals)))
return(NULL)
print(plot_roc(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_roc,
yaxisrange = input$yrange_roc,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.roc.df.rdata <- downloadHandler(
filename = "roc-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.roc.df.tsv <- downloadHandler(
filename = "roc-data.tsv",
content = function(file) {
roc_df <- isolate(roc(plotvalues()$all_vals))
utils::write.table(roc_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$roc <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(roc(plotvalues()$all_vals)),
paste0("ROC curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(roc(plotvalues()$all_vals)))
return(NULL)
plot_roc(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
xaxisrange = input$xrange_roc, yaxisrange = input$yrange_roc,
linewidth = 1)
})
})
output$roc_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(roc(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(roc(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$roc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FPR", yvar = "TPR")
} else {
res <- nearPoints(all_data, input$roc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FPR", yvar = "TPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("FPR", "TPR"),
tabtype = "small")
})
## -------------------------- SCATTER ------------------------------ ##
output$plot.scatter <- renderUI({
plotOutput("scatter", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "scatter_plot_click")
})
## Figure for exporting
output$export.scatter <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(scatter(plotvalues()$all_vals)))
return(NULL)
print(plot_scatter(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, doflip = input$doflip,
dolog = input$dolog))
grDevices::dev.off()
})
## Data for exporting
output$export.scatter.df.rdata <- downloadHandler(
filename = "scatter-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.scatter.df.tsv <- downloadHandler(
filename = "scatter-data.tsv",
content = function(file) {
scatter_df <- isolate(scatter(plotvalues()$all_vals))
utils::write.table(scatter_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$scatter <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(scatter(plotvalues()$all_vals)),
paste0("Scatter plots can not be displayed. Check that ",
"cont_truth is not 'none' and that scores ",
"are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(scatter(plotvalues()$all_vals)))
return(NULL)
plot_scatter(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize, doflip = input$doflip,
dolog = input$dolog)
})
})
output$scatter_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(scatter(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(scatter(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$scatter_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "OBSERVATION",
yvar = "TRUTH")
} else {
res <- nearPoints(all_data, input$scatter_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "OBSERVATION", yvar = "TRUTH")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("observation", "truth"),
tabtype = "scatter")
})
## ---------------------------- FPC -------------------------------- ##
output$plot.fpc <- renderUI({
plotOutput("fpc", width = "100%", height = paste0(input$plotheight, "px"),
hover = "fpc_plot_click")
})
## Figure for exporting
output$export.fpc <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpc(plotvalues()$all_vals)))
return(NULL)
print(plot_fpc(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
maxnfdc = input$maxnfdc,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fpc.df.rdata <- downloadHandler(
filename = "fpc-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$export.fpc.df.tsv <- downloadHandler(
filename = "fpc-data.tsv",
content = function(file) {
fpc_df <- isolate(fpc(plotvalues()$all_vals))
utils::write.table(fpc_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fpc <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(
need(is_plottable(fpc(plotvalues()$all_vals)),
paste0("False discovery curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpc(plotvalues()$all_vals)))
return(NULL)
plot_fpc(cobraplot = plotvalues()$all_vals, title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
maxnfdc = input$maxnfdc,
linewidth = 1)
})
})
output$fpc_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fpc(plotvalues()$all_vals)))
return(NULL)
all_data <- isolate(fpc(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(all_data, input$fpc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "topN", yvar = "FP")
} else {
res <- nearPoints(all_data, input$fpc_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "topN", yvar = "FP")
}
fix_res(res, methodcol = "fullmethod",
aspcts = c("Number of detections", "FP"), tabtype = "small")
})
## -------------------------- FDRTPR ------------------------------- ##
output$plot.fdrtprcurve <- renderUI({
plotOutput("fdrtprcurve", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "fdrtprcurve_plot_click")
})
## Figure for exporting
output$export.fdrtprcurve <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fdrtprcurve(plotvalues()$all_vals)))
return(NULL)
print(plot_fdrtprcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrtpr,
yaxisrange = input$yrange_fdrtpr,
plottype = input$plottype,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fdrtpr.df.rdata <- downloadHandler(
filename = "fdrtpr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$fdrtpr.df.tsv.button <- renderUI({
if ("curve" %in% input$plottype)
downloadButton("export.fdrtpr.df.tsv", label = "Download tsv (curve)")
else
downloadButton("export.fdrtpr.df.tsv", label = "Download tsv (points)")
})
output$export.fdrtpr.df.tsv <- downloadHandler(
filename = "fdrtpr-data.tsv",
content = function(file) {
fdrtpr_curve_df <- isolate(fdrtprcurve(plotvalues()$all_vals))
fdrtpr_point_df <- isolate(fdrtpr(plotvalues()$all_vals))
if ("curve" %in% input$plottype)
utils::write.table(fdrtpr_curve_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
else if ("points" %in% input$plottype)
utils::write.table(fdrtpr_point_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fdrtprcurve <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(any(c("points", "curve") %in% input$plottype),
"No plot type selected.")
)
if ("curve" %in% input$plottype)
validate(
need(is_plottable(fdrtprcurve(plotvalues()$all_vals)),
paste0("FDR/TPR curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
if ("points" %in% input$plottype)
validate(
need(is_plottable(fdrtpr(plotvalues()$all_vals)),
paste0("FDR/TPR at specified adjusted p-value thresholds ",
"can not be displayed. Check that binary_truth is not ",
"'none' and that adjusted p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrtprcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrtpr(plotvalues()$all_vals)))
return(NULL)
if (!any(c("curve", "points") %in% input$plottype))
return(NULL)
plot_fdrtprcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrtpr,
yaxisrange = input$yrange_fdrtpr,
plottype = input$plottype,
linewidth = 1)
})
})
output$fdrtprcurve_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrtprcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrtpr(plotvalues()$all_vals)))
return(NULL)
if ("curve" %in% input$plottype)
tab_data <- isolate(fdrtprcurve(plotvalues()$all_vals))
else tab_data <- isolate(fdrtpr(plotvalues()$all_vals))
if ("split" %in% isolate(input$facet_opt)) {
res <- nearPoints(tab_data, input$fdrtprcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FDR", yvar = "TPR")
} else {
res <- nearPoints(tab_data, input$fdrtprcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FDR", yvar = "TPR")
}
fix_res(res, methodcol = "fullmethod", aspcts = c("FDR", "TPR"),
tabtype = "large")
})
## -------------------------- FDRNBR ------------------------------- ##
output$plot.fdrnbrcurve <- renderUI({
plotOutput("fdrnbrcurve", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "fdrnbrcurve_plot_click")
})
## Figure for exporting
output$export.fdrnbrcurve <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fdrnbrcurve(plotvalues()$all_vals)))
return(NULL)
print(plot_fdrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrnbr,
plottype = input$plottype,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fdrnbr.df.rdata <- downloadHandler(
filename = "fdrnbr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$fdrnbr.df.tsv.button <- renderUI({
if ("curve" %in% input$plottype)
downloadButton("export.fdrnbr.df.tsv", label = "Download tsv (curve)")
else
downloadButton("export.fdrnbr.df.tsv", label = "Download tsv (points)")
})
output$export.fdrnbr.df.tsv <- downloadHandler(
filename = "fdrnbr-data.tsv",
content = function(file) {
fdrnbr_curve_df <- isolate(fdrnbrcurve(plotvalues()$all_vals))
fdrnbr_point_df <- isolate(fdrnbr(plotvalues()$all_vals))
if ("curve" %in% input$plottype)
utils::write.table(fdrnbr_curve_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
else if ("points" %in% input$plottype)
utils::write.table(fdrnbr_point_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fdrnbrcurve <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(any(c("points", "curve") %in% input$plottype),
"No plot type selected.")
)
if ("curve" %in% input$plottype)
validate(
need(is_plottable(fdrnbrcurve(plotvalues()$all_vals)),
paste0("FDR/NBR curves can not be displayed. Check that ",
"binary_truth is not 'none'."))
)
if ("points" %in% input$plottype)
validate(
need(is_plottable(fdrnbr(plotvalues()$all_vals)),
paste0("FDR/NBR at specified adjusted p-value thresholds ",
"can not be displayed. Check that binary_truth is not ",
"'none' and that adjusted p-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrnbr(plotvalues()$all_vals)))
return(NULL)
if (!any(c("curve", "points") %in% input$plottype))
return(NULL)
plot_fdrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fdrnbr,
plottype = input$plottype,
linewidth = 1)
})
})
output$fdrnbrcurve_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fdrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fdrnbr(plotvalues()$all_vals)))
return(NULL)
if ("curve" %in% input$plottype)
tab_data <- isolate(fdrnbrcurve(plotvalues()$all_vals))
else tab_data <- isolate(fdrnbr(plotvalues()$all_vals))
if ("split" %in% input$facet_opt) {
res <- nearPoints(tab_data, input$fdrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FDR", yvar = "NBR")
} else {
res <- nearPoints(tab_data, input$fdrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FDR", yvar = "NBR")
}
methodcol <- "fullmethod"
fix_res(res, methodcol = methodcol,
aspcts = c("FDR", "Number of detections"), tabtype = "large")
})
## -------------------------- FSRNBR ------------------------------- ##
output$plot.fsrnbrcurve <- renderUI({
plotOutput("fsrnbrcurve", width = "100%",
height = paste0(input$plotheight, "px"),
hover = "fsrnbrcurve_plot_click")
})
## Figure for exporting
output$export.fsrnbrcurve <- downloadHandler(
filename = "shiny-plot.pdf",
content = function(file) {
grDevices::pdf(file, width = 12, height = input$plotheight/67)
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)) ||
!is_plottable(fsrnbrcurve(plotvalues()$all_vals)))
return(NULL)
print(plot_fsrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fsrnbr,
plottype = input$plottype,
linewidth = 1))
grDevices::dev.off()
})
## Data for exporting
output$export.fsrnbr.df.rdata <- downloadHandler(
filename = "fsrnbr-data.Rdata",
content = function(file) {
cobraplot <- isolate(plotvalues()$all_vals)
save(cobraplot, file = file)
})
output$fsrnbr.df.tsv.button <- renderUI({
if ("curve" %in% input$plottype)
downloadButton("export.fsrnbr.df.tsv", label = "Download tsv (curve)")
else
downloadButton("export.fsrnbr.df.tsv", label = "Download tsv (points)")
})
output$export.fsrnbr.df.tsv <- downloadHandler(
filename = "fsrnbr-data.tsv",
content = function(file) {
fsrnbr_curve_df <- isolate(fsrnbrcurve(plotvalues()$all_vals))
fsrnbr_point_df <- isolate(fsrnbr(plotvalues()$all_vals))
if ("curve" %in% input$plottype)
utils::write.table(fsrnbr_curve_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
else if ("points" %in% input$plottype)
utils::write.table(fsrnbr_point_df, file = file, quote = FALSE,
row.names = FALSE, col.names = TRUE, sep = "\t")
})
output$fsrnbrcurve <- renderPlot({
## Generate error messages if data not available.
validate(
need(length(values$all_methods) > 0 & length(input$cols) > 0 &
(input$goButton > 0 | isTRUE(autorun)),
paste0("No input data provided, or 'Start calculation!' has ",
"not been clicked in order to launch calculations."))
)
validate(need(any(c("points", "curve") %in% input$plottype),
"No plot type selected.")
)
if ("curve" %in% input$plottype)
validate(
need(is_plottable(fsrnbrcurve(plotvalues()$all_vals)),
paste0("FSR/NBR curves can not be displayed. Check that ",
"cont_truth is not 'none' and that s-values ",
"are provided."))
)
if ("points" %in% input$plottype)
validate(
need(is_plottable(fsrnbr(plotvalues()$all_vals)),
paste0("FSR/NBR at specified adjusted p-value thresholds ",
"can not be displayed. Check that cont_truth is not ",
"'none' and that s-values are provided."))
)
withProgress(message = "Updating plot...", value = 0, {
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fsrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fsrnbr(plotvalues()$all_vals)))
return(NULL)
if (!any(c("curve", "points") %in% input$plottype))
return(NULL)
plot_fsrnbrcurve(cobraplot = plotvalues()$all_vals,
title = plotvalues()$title,
stripsize = input$stripsize, titlecol = "white",
pointsize = input$pointsize,
xaxisrange = input$xrange_fsrnbr,
plottype = input$plottype,
linewidth = 1)
})
})
output$fsrnbrcurve_click_info <- DT::renderDataTable({
if (length(values$all_methods) == 0 | length(input$cols) == 0 |
(input$goButton == 0 & !isTRUE(autorun)))
return(NULL)
if ("curve" %in% input$plottype &
!is_plottable(fsrnbrcurve(plotvalues()$all_vals)))
return(NULL)
if ("points" %in% input$plottype &
!is_plottable(fsrnbr(plotvalues()$all_vals)))
return(NULL)
if ("curve" %in% input$plottype)
tab_data <- isolate(fsrnbrcurve(plotvalues()$all_vals))
else tab_data <- isolate(fsrnbr(plotvalues()$all_vals))
if ("split" %in% input$facet_opt) {
res <- nearPoints(tab_data, input$fsrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, panelvar1 = "splitval",
xvar = "FSR", yvar = "NBR")
} else {
res <- nearPoints(tab_data, input$fsrnbrcurve_plot_click,
threshold = 5, maxpoints = 100,
addDist = TRUE, xvar = "FSR", yvar = "NBR")
}
methodcol <- "fullmethod"
fix_res(res, methodcol = methodcol,
aspcts = c("FSR", "Number of detections"), tabtype = "fsrnbr")
})
## Close app
shiny::observeEvent(input$close_app, {
shiny::stopApp()
})
} ## end of server_function
#nocov end
shinyApp(ui = p_layout, server = server_function)
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.