Embed the linked display from detourr in the pandemonium app

[Divendra2006]

This PR introduces linked 2D tour visualizations in the pandemonium Shiny app using the detourr package. The changes add interactive scatterplot animations to explore projections of the dataset.

Changes Made:

• Updated ui.R:

Integrated bscols() to display two linked displayScatter2dOutput() components for interactive visualizations.

• Updated server.R:

Created a shared dataset SharedData$new(as.data.frame(pred)) for linking selections across visualizations.

Implemented shinyRenderDisplayScatter2d() for two different tour paths:

• tour1 uses grand_tour(2)
• tour2 uses little_tour(2)

Purpose of This Change:

• Adds interactive 2D tour visualizations using detourr
• Implements linking and brushing with crosstalk for better data exploration.

Add this feature at the navbar

[Divendra2006]

@uschiLaa @gevalencia ,
I complete the hard test of pandemonium , can you review this and suggest any improvement.

[uschiLaa]

Hi @Divendra2006, I have tried and it runs well for me. However, it seems there is an issue with the color assignment, which does not match the colors in other displays in the app. Maybe you can check why this does not work properly.

[Divendra2006]

Hi @uschiLaa,
Thank you for reviewing the code and pointing out the issue with the color assignment . I checked and find out that this happen because the colour aesthetic in detourr is mapped to rv$groups (numeric cluster IDs) instead of the actual color values stored in rv$col.
But changing into rv$col gives a error .

can you guide me to solve this problem?

[Divendra2006]

Hi @uschiLaa ,
To solve color assignment problem , when i use detourr::detourr() with tour_aes(colour = ~colour) , I reciece the following error by running this code

shiny::observeEvent(rv$groups, {
    shared_data <- crosstalk::SharedData$new(cbind(as.data.frame(pred), colour = as.factor(rv$col)))

    output$tour1 <- detourr::shinyRenderDisplayScatter2d({
        detourr::detour(shared_data, detourr::tour_aes(projection = -1, colour = ~colour)) |>
            detourr::tour_path(tourr::grand_tour(2), fps = 60) |>
            detourr::show_scatter(alpha = 0.7, axes = FALSE)
    })

    output$tour2 <- detourr::shinyRenderDisplayScatter2d({
        detourr::detour(shared_data, detourr::tour_aes(projection = -1, colour = ~colour)) |>
            detourr::tour_path(tourr::little_tour(2), fps = 60) |>
            detourr::show_scatter(alpha = 0.7, axes = FALSE)
    })
})

Error in purrr::map: ℹ In index: 2.ℹ With name: colour.Caused by error in `.f()`:! Formula shorthand must be wrapped in `where()`.  # Bad  data %>% select(~colour)  # Good  data %>% select(where(~colour))

This error suggests that detourr is internally using tidyselect functions (likely select()) in a way that's not compatible with the current tidyselect behavior.

[uschiLaa]

Can you try without the "~" for the colour argument? You might also need to pass the color palette to detourr.

[Divendra2006]

Thank you for your suggestions,
Unfortunately, removing the ~ and adding the palette argument did not resolve the issue. I am still recieving the "Can't select columns that don't exist."
this is updated code

shiny::observeEvent(rv$groups, {
    shared_data <- crosstalk::SharedData$new(cbind(as.data.frame(pred), colour = rv$col))

    output$tour1 <- detourr::shinyRenderDisplayScatter2d({
        detourr::detour(shared_data, detourr::tour_aes(projection = -1, colour = rv$col, palette = rv$pal)) |>
            detourr::tour_path(tourr::grand_tour(2), fps = 60) |>
            detourr::show_scatter(alpha = 0.7, axes = FALSE)
    })

    output$tour2 <- detourr::shinyRenderDisplayScatter2d({
        detourr::detour(shared_data, detourr::tour_aes(projection = -1, colour = rv$col, palette = rv$pal)) |>
            detourr::tour_path(tourr::little_tour(2), fps = 60) |>
            detourr::show_scatter(alpha = 0.7, axes = FALSE)
    })
})

this is error that i encounter

Caused by error in `.f()`:
! Can't select columns that don't exist.
✖ Columns `#D95F02`, `#D95F02`, `#D95F02`, `#D95F02`, `#D95F02`, etc. don't exist.

[uschiLaa]

What I mean is using the column from the new version of the data frame you have created, but without the ~ since I think you can directly specify the column name. With the latest version you are trying to provide a vector for the color argument.

[Divendra2006]

Thank you for pointing this ,
But the same error come when i use "colour" as the column name

shiny::observeEvent(rv$groups, {
    shared_data <- crosstalk::SharedData$new(cbind(as.data.frame(pred), colour = rv$col))

    output$tour1 <- detourr::shinyRenderDisplayScatter2d({
        detourr::detour(shared_data, detourr::tour_aes(projection = -1, colour = "colour", palette = rv$pal)) |>
            detourr::tour_path(tourr::grand_tour(2), fps = 60) |>
            detourr::show_scatter(alpha = 0.7, axes = FALSE)
    })

    output$tour2 <- detourr::shinyRenderDisplayScatter2d({
        detourr::detour(shared_data, detourr::tour_aes(projection = -1, colour = "colour", palette = rv$pal)) |>
            detourr::tour_path(tourr::little_tour(2), fps = 60) |>
            detourr::show_scatter(alpha = 0.7, axes = FALSE)
    })
})

the error comes

 Can't select columns that don't exist.
✖ Columns `#1B9E77`, `#D95F02`, and `#7570B3` don't exist.

I think the problem is in the detourr package (package's color maping logic) that misinterprets the "colour" column in data frame . It's is designed to use the column name as a reference ,but it's incorrectly using the colour values themselves as coloumn name.