How to convert a piece of R code into Shiny apps? - r

Question

A new self-learner for Shiny apps, still trying to explore the structure of Shiny apps. I have a piece of code which I want to convert into Shiny apps. I wonder if someone can walk me through how the process goes. My goal is to make the figure title/subtitle dynamic based on inputs (i.e., lnHR0, p_ctl0, tfu) to function. Thanks!!!
R code:
library(tidyverse)
library(ggplot2)
WR_sim_OC <- function(n_trt, n_ctl, lnHR0, p_trt0, med_ctl, tfu, enroll, dur_boost){
#n_trt=60: number of subjects (treatment)
#n_ctl=30: number of subjects (control)
#lnHR0=log(1): Overall Survival Log Hazard Ratio
#p_trt0 = seq(0.46, 0.66, 0.01): Response Rate (Treatment)
#med_ctl=21.8: Median Overall Survival (Control)
#tfu=9: Minimum Follow-up Time
#enroll=19: Enrollment Time
#dur_boost=0: Durability Boost (percentage)
## insert real simulation code here ##
## Fake Results
results <- tibble(ORR_trt = p_trt0, avg_HR = rep(0.774, times = 4), maturity = rep(36.7, times = 4), ORR = c(20.71, 38.87, 60.61, 78.95),
WR = c(46.8, 56.0, 64.3, 72.8), WO = c(46.0, 55.0, 63.7, 71.9), OS = rep(55.0, times = 4))
## Create Operating Characteristic Figure
dat <- results %>% pivot_longer(c(ORR, WR, WO, OS), names_to = 'Method', values_to = 'POS')
out1 <- ggplot(data = dat, aes(x = ORR_trt, y = POS, group = Method)) +
geom_line(aes(color = Method), size = 1) +
geom_point(aes(color = Method), size = 2.5) +
theme(legend.position = 'bottom') +
labs(title = 'HR=1.0 Treatment vs. Control', subtitle = 'ORR in Control Arm=46%, 9mo follow-up', color = 'Method') +
ylab('Probability of Incorrect Go') +
xlab('ORR in Treatment Arm') +
ylim(0, 100)
## Output OC Table + Figure to shiny app
list(results, out1)
}
WR_sim_OC(n_trt=60, n_ctl=30, lnHR0=log(1), p_ctl0=0.46, p_trt0 = seq(0.46, 0.66, 0.06), med_ctl=21.8, tfu=9, enroll=19, dur_boost=0)
I tried writing the ui.R as follows (suppose the suv_plot is the output name), which I know is wrong. The server.R part is too hard for me... Can someone help?
fluidPage(
numericInput("lnHRO",
label = h3("ln(HRO)"),
value = log(1)),
numericInput("pctl",
label = h3("Response Rate (Control)"),
value = 0.46),
numericInput("tfu",
label = h3("Minimum Follow-up Time (Month)"),
value = 9),
hr(),
plotOutput("suv_plot")
)

Answer 1

My first suggestion is just to look at tutorials on shiny, they give a great overview on how to start a project: https://shiny.rstudio.com/tutorial/
I didn't know a thing about programming a few years back, so I understand it can be hard figuring out where to start, so I wanted to give you an idea of how to implement a function, and use shiny inputs to make the resulting table/plot be dynamic.
I switched up your code to be easier to reproduce for myself. I hope this gives you the starting point you need:
library(tidyverse)
library(ggplot2)
library(shiny)
WR_sim_OC <- function(MPG, CYL, DISP){
results <- mtcars%>% #Function to make a table
filter(cyl > CYL,
mpg > MPG,
disp > DISP)
out1 <- ggplot(data = results, aes(x = mpg, y = disp, group = cyl)) +
geom_line(aes(color = hp), size = 1) #Function to make a plot
list(results, out1) #List to create table and function
}
ui <- fluidPage(
numericInput("MilesPerGallon", "mpg", value = 15),
numericInput("Cylinders", "cyl", value = 4),
numericInput("Displacement", "disp", value = 200),
tableOutput("TABLE"),
plotOutput("PLOT")
)
server <- function(input, output, session) {
output$TABLE<-renderTable({
req(input$MilesPerGallon, input$Cylinders, input$Displacement) #Requires all three inputs before it makes the table
WR_sim_OC(input$MilesPerGallon, input$Cylinders, input$Displacement)[1] #Only pulling the table from the function
})
output$PLOT<-renderPlot({
req(input$MilesPerGallon, input$Cylinders, input$Displacement) #Requires all three inputs before it makes the plot
WR_sim_OC(input$MilesPerGallon, input$Cylinders, input$Displacement)[2] #Only pulling the plot from the function
})
}
shinyApp(ui, server)
Essentially on the server side where you render the plot or table, you use those inputs from the ui as the dynamic points in your function. I used req() for both of the renderTable and renderPlot to make sure the inputs are filled out before it makes the table plot. Best of luck!