Ggplot: plot - interaction - exclude --> prevent excluded points reset - r

Question

I would like to prevent reseting of the curve while choosing second variable to display with selectizeInput. For example in the code below we choose one value (mtcars dataset) in selectizeInput of cyl (6), and exclude one point from the curve, then we choose second value of cyl (4) to display, and therefore the previous curve with cyl=6, resets itself (the point which has been excluded, appears again).
Is there any way to prevent this behaviour and while choosing second variable the excluded point stays "excluded"?
Example code:
library(ggplot2)
library(shiny)
ui <- fluidPage(
fluidRow(
column(width = 6,
plotOutput("plot1", height = 350,
click = "plot1_click"),
selectizeInput("valuecyl", "Select value of cyl:", choices=unique(mtcars$cyl), multiple = TRUE))
)
)
server <- function(input, output) {
# For storing which rows have been excluded
vals <- reactiveValues()
data_df <- reactive({
data <- mtcars
data <- data[data$cyl %in% input$valuecyl, ]
vals$keeprows = rep(TRUE, nrow(data))
data
})
output$plot1 <- renderPlot({
data<- data_df()
keep <- data[ vals$keeprows, , drop = FALSE]
exclude <- data[!vals$keeprows, , drop = FALSE]
print(keep)
ggplot(keep, aes(wt,mpg,colour=as.factor(cyl))) + geom_point(data=keep) + geom_line(data=keep) +
geom_point(data = exclude, shape = 21, fill = NA, color = "black", alpha = 0.25)
})
# Toggle points that are clicked
observeEvent(input$plot1_click, {
data <- data_df()
res <- nearPoints(data, input$plot1_click, allRows = TRUE)
vals$keeprows <- xor(vals$keeprows, res$selected_)
})
}
shinyApp(ui, server)

Answer 1

The issue here is that you are overwriting vals$keeprows and replacing it with rep(TRUE, nrow(data)) every time the user selects a value in your selectize.
You need to update the keeprows variable by keeping the rows that were kept by the user and adding in the new rows coming from the extra selection.
To do that, I slightly modified your code:
#added the data in the reactiveValues for convenience
vals <- reactiveValues(keeprows=logical(0),data=mtcars[0,])
#this observes the input and updates the data when the user adds a cyl value
observeEvent(input$valuecyl,{
#get the id (here rownames) of the points excluded by the user
excluded_ids <- rownames(vals$data)[!vals$keeprows]
#make the new data
vals$data=mtcars[mtcars$cyl %in% input$valuecyl,]
#keep the rows that the user had not previously excluded.
vals$keeprows = !(rownames(vals$data) %in% excluded_ids)
})
Since I added the data in the reactiveValues and removed data_df, you need to replace data_df() by vals$data in your code for this to work.