Creating several selectInputs from same variable in Shiny & calling in plot - r

Question

I am challenged trying to implement a feature into my Shiny app. The problem is two-fold:
Is it possible to have 2 inputs from the same variable? I have one variable that is a list of indicators. I want the user to be able to select 2 indicators with selectInput, and then draw a scatter plot. There has to be 2 selectInputs because other parts of the app will rely on only the first selectInput. My data is long. I don't think it will work if I make it wide because my data includes latitude and longitude information so it wouldn't make sense to create a selectInput with names(data), for example.
If I can have 2 selectInputs from the same variable, how would I call the values in my plot, since the value is called 'value' for both the inputs?
EDIT: Following Gregor's suggestion to reference the inputs with aes_string, I would expect the following example of mtcars gathered into long format to work, but I instead get an aesthetics or object not found error. I think I probably need to filter the data, but I don't understand how I can do that since my variable indicators now refers to both 'indicators' and 'indicators2'. I.e., I can't have
filtered <-
cars %>%
filter(indicators == input$indicators,
indicators == input$indicators2)
Maybe I need to create a reactive expression that creates a new data frame instead? This is my non-working reproducible code with long-form mtcars:
library(ggplot2)
library(shiny)
cars <- mtcars %>%
gather(indicators, value, mpg:carb)
ui <- fluidPage(
# Application title
titlePanel("mtcars"),
sidebarLayout(
sidebarPanel(
selectInput("indicators",
label = "select indicator:",
choices = c("mpg", "cyl", "disp", "hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb")
),
selectInput("indicators2",
label = "select indicator:",
choices = c("mpg", "cyl", "disp", "hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb")
)
),
mainPanel(
plotOutput("carsPlot")
)
)
)
server <- function(input, output) {
output$carsPlot <- renderPlot({
ggplot(cars, aes_string(x = input$indicators, y = input$indicators2)) +
geom_point(shape = 1)
})
}
# Run the application
shinyApp(ui = ui, server = server)

Answer 1

You can use aes_string to pass input$indicators and input$indicators2 to ggplot like this. There is no need to cast your data into wide format since ggplot can actually handle long data better.
library(ggplot2)
library(shiny)
ui <- fluidPage(
# Application title
titlePanel("mtcars"),
sidebarLayout(
sidebarPanel(
selectInput("indicators",
label = "select indicator:",
choices = names(mtcars)),
selectInput("indicators2",
label = "select indicator:",
choices = names(mtcars))
),
mainPanel(
plotOutput("carsPlot")
)
)
)
server <- function(input, output) {
output$carsPlot <- renderPlot({
ggplot(mtcars, aes_string(x = input$indicators, y = input$indicators2)) +
geom_point(shape = 1)
})
}
# Run the application
shinyApp(ui = ui, server = server)

Answer 2

Here another solution with a selectInput in multiple mode.
library(dplyr)
library(tidyr)
library(ggplot2)
library(shiny)
cars <- mtcars %>%
gather(indicators, value, mpg:carb)
ui <- fluidPage(
titlePanel("mtcars"),
sidebarLayout(
sidebarPanel(
uiOutput("ui_indicators")
),
mainPanel(
plotOutput("carsPlot")
)
)
)
server <- function(input, output) {
output$ui_indicators <- renderUI({
choices <- unique(cars$indicators)
selectInput("indicators",
label = "select indicators :",
choices = choices,
multiple = TRUE)
})
output$carsPlot <- renderPlot({
filtered <- cars %>% filter(indicators %in% input$indicators)
ggplot(filtered, aes(x = indicators, y = value)) +
geom_point(shape=1)
})
}
# Run the application
shinyApp(ui = ui, server = server)

Answer 3

Building on the tip from Gregor on aes_string, I managed to fix this. I ended up using the wide data and adding a proper reactive statement that creates a new data frame out of the selected indicators.
My server function now looks like this:
server <- function(input, output) {
selectedVars <- reactive({
cars[, c(input$indicators, input$indicators2)]
})
output$carsPlot <- renderPlot({
ggplot(selectedVars(), aes_string(x = input$indicators, y = input$indicators2)) +
geom_point(shape = 1)
})
}
All works beautifully, and I am beginning to learn more about the utility of reactive functions in Shiny :)
Gregor de Cillia provided the answer I was looking for.
The two inputs (which are not a problem) can be referenced using aes_string.