I use ggsurvplot to draw a survival curve, and I want to input text to the parameter P value. When the input content is character, it can be displayed correctly, however, when the input content is numeric, an error will occur.

The input data is as follows：

enter image description here

The full code is as follows:

rm(list = ls())
options(scipen = 200)
options(encoding = "UTF-8")
options(stringsAsFactors = TRUE)
library(survival)
library(survminer)
library(shiny)
library(bslib)
library(shinythemes)
mIHC <<- read.csv("0 expr.csv",header=TRUE,row.names=1,check.names = FALSE)
gene_list <<- colnames(mIHC)[3: dim(mIHC)[2]]
gene_list_order = gene_list[order(gene_list)]

ui <- fixedPage(
  tags$style(HTML("
          .navbar .navbar-header {float: left}
          .navbar .navbar-nav {float: right}

        ")
  ),
  navbarPage(
    windowTitle = "GMAP",
    fluid = TRUE,
    # theme = bs_theme(bootswatch = "flatly",),
    title = span("GMAP"),
    tabPanel(
      "Introduction",
    ),
    tabPanel(
      "Survival analysis",
      sidebarLayout(
          sidebarPanel(width = 5,
                       selectInput("gene_name", "Gene symbol", choices = gene_list_order),
                       sliderInput("cutoff_per", "Cutoff percent", 
                                   value = 0.5, min = 0, max = 0.99, step = 0.01,
                                   ticks = TRUE)
                       ),
        mainPanel(width = 7,
                  tabsetPanel(
                    tabPanel("Plot", 
                             plotOutput("surv", width = "420px", height = "400px"),
                             downloadButton('downloadPlot','Download Plot')),
                    tabPanel("Summary"),
                    tabPanel("Table")
                   )
                  )
      )
    ),
    tabPanel(
      "Statistics analysis",
    ),
    tabPanel(
      "Heatmap"
    ),
    tabPanel(
      "About"
    )
  )
)



server <- function(input, output, session) {
  env <- parent.frame()
  plot2 <- reactive({
    gene_name = input$gene_name
    cutoff_per = input$cutoff_per
    surv_gene = mIHC[ , c("OS", "event", gene_name)]
    
    plot(surv_gene$OS, surv_gene$event)
  })
  
  surv_plot <- reactive({
    gene_name = input$gene_name
    cutoff_per = input$cutoff_per
    surv_gene = mIHC[ , c("OS", "event", gene_name)]
    surv_temp = surv_gene
    surv_temp = cbind(surv_temp,surv_temp[,1])
    colnames(surv_temp) = c("OS", "event", gene_name, "group")
    for (row_place in 1: dim(surv_temp)[1]) {
      if(surv_temp[row_place, 3] > quantile(surv_temp[,3], cutoff_per)) {
        surv_temp[row_place, "group"] = "high"
      } else {
        surv_temp[row_place, "group"] = "low"
      }
    }
    surv_gene <- surv_temp
    

    fit <- eval(parse(text = paste0("survfit(Surv(OS, event) ~ group, data = surv_gene)")))
    p_val = surv_pvalue(fit, data = surv_gene, method = "1")
    p_val = round(as.numeric(p_val),2)
    # p_val = as.character(p_val)
    # p_val = "abc"
    
    ggsurv_doc <- eval(parse(text = paste0("survfit(Surv(OS, event) ~ group, data = surv_gene)")))
    
    ggpar(
    ggsurvplot(ggsurv_doc,
               data = surv_gene,
               # ggtheme = theme_bw(),
               conf.int = F,
               censor = T,
               palette = c("#DC143C", "#4071B3"),
               legend.title = colnames(surv_gene)[3],
               pval = paste("P =", p_val),
               # pval = T,
               legend.labs=c("High", "Low"),
               # legend.labs=unique(surv_gene$group),
               surv.median.line = "hv",
               break.time.by = 12,
               xlab = "Time (months)",
    ),
    font.main = 13,
    font.submain = 13,
    font.x = 13,
    font.y = 13,
    font.caption = 13,
    font.title = 13,
    font.subtitle = 13,
    font.legend = 13,
    font.tickslab = 13,
    )
  })

  output$surv <- renderPlot({
    surv_plot()
  }, res = 96)
  
  output$downloadPlot <- downloadHandler(
    filename = function() {
      paste("plot.pdf")
    },
    content = function(file) {
     pdf(file, width = 4.5,height = 4.5)
     print(surv_plot(), newpage = FALSE)
     dev.off()
    }
  )
}

shinyApp(ui, server)

<iframe name="sif1" sandbox="allow-forms allow-modals allow-scripts" frameborder="0"></iframe> enter image description here

It works correctly when p_val is a character, as follow:

    p_val = "abc"

<iframe name="sif2" sandbox="allow-forms allow-modals allow-scripts" frameborder="0"></iframe> enter image description here

CodePudding user response：

This looks wrong:

p_val = surv_pvalue(fit, data = surv_gene, method = "1")
p_val = round(as.numeric(p_val),2)

surv_pvalue returns a data.frame but you seem to be treating it like a numeric. Perhaps try:

 p_val = surv_pvalue(fit, data = surv_gene, method = "1")$pval[[1]]
 p_val = round(as.numeric(p_val),2)