R Shiny ggplot2 line chart won't show lines when using "key" property and facet_grid

Question

I have an R shiny app that is displaying a line chart with a facet grid. I therefore use ggplot2 and plotly.

Now I want to make it possible to click a position on a line and retrieve the corresponding data row from the data frame.

I learned fetching the click event is possible by catching the "plotly_click" click event via event_data("plotly_click") . This works; I get a curve number, x- and y- coordinates. Now to get a reference to my data row I learned I have to use a "key" property for ggplot like stated here: How can I grab the row of data from a ggplotly in shiny .

Now when I add the "key" property to my ggplot suddenly the lines won't show up anymore (it seems the chart stays empty). Interestingly, when I remove facet_grid, some lines will show up and clicking it provides the "key"-information with the event like stated in the link above.

EDIT:

I reproduced the behavior with the mtcars exmaple:

library(shiny)
library(plotly)
library(tidyr)

mtcars$key <- row.names(mtcars)

ui <- fluidPage(
    plotlyOutput("originalLinePlot"),
    plotlyOutput("keyLinePlot"),
    verbatimTextOutput("click"),
)

server <- function(input, output) {

    output$originalLinePlot <- renderPlotly({
        # here I want to add click event with data row selection - click doesn't return key info

        data_long <- gather(mtcars, condition, measurement, c(drat, wt), factor_key=TRUE)

        g <- ggplot(data_long, aes(x=mpg))
        # won't work
        # g <- ggplot(data_long, aes(x=mpg, key=key))

        g <- g + facet_grid(rows = vars(condition), scales="free_y")
        g <- g + geom_line(aes(y=measurement))
        g
    })


    output$keyLinePlot <- renderPlotly({
        data_long <- gather(mtcars, condition, measurement, c(drat, wt), factor_key=TRUE)

        g <- ggplot(data_long, aes(x=mpg, key=key))

        # won't work
        # g <- g + facet_grid(rows = vars(condition), scales="free_y")
        g <- g + geom_line(aes(y=measurement))
        g
    })

    output$click <- renderPrint({
        d <- event_data("plotly_click")
        if (is.null(d)) "Click events appear here (double-click to clear)" else data.frame(d) 
    })
}

shinyApp(ui = ui, server = server)

May it be that due to the key-property the line chart gets confused as how to draw its lines? The lines appearing when removing the facet_grid and having the "key"-propery included look quite strange..

Any ideas how to resolve this? Can I solve my issue also in a different way than using the key-property?

Thanks and BR!

Answer 1

I've found a solution of how to solve this: by combining points and lines and adding the key information to the points instead of the lines - see second plot:

library(shiny)
library(plotly)
library(tidyr)

mtcars$key <- row.names(mtcars)

ui <- fluidPage(
    plotlyOutput("originalLinePlot"),
    plotlyOutput("keyLinePlot"),
    verbatimTextOutput("click"),
)

server <- function(input, output) {

    output$originalLinePlot <- renderPlotly({
        # here I want to add click event with data row selection - click doesn't return key info

        data_long <- gather(mtcars, condition, measurement, c(drat, wt), factor_key=TRUE)

        g <- ggplot(data_long, aes(x=mpg))
        # won't work
        # g <- ggplot(data_long, aes(x=mpg, key=key))

        g <- g + facet_grid(rows = vars(condition), scales="free_y")
        g <- g + geom_line(aes(y=measurement))
        g
    })


    output$keyLinePlot <- renderPlotly({
        data_long <- gather(mtcars, condition, measurement, c(drat, wt), factor_key=TRUE)

        g <- ggplot(data_long, aes(x=mpg))

        g <- g + facet_grid(rows = vars(condition), scales="free_y")
        g <- g + geom_line(aes(y=measurement))
        g <- g + geom_point(aes(y=measurement, key=key))
        g
    })

    output$click <- renderPrint({
        d <- event_data("plotly_click")
        if (is.null(d)) "Click events appear here (double-click to clear)" else data.frame(d) 
    })
}

shinyApp(ui = ui, server = server)