gather nodes based on group information

shiny minimal example

require(shiny)
require(visNetwork)

server <- function(input, output) {
  output$network <- visNetwork::renderVisNetwork({
    edges <- data.frame(
      from = sample(1:10, 8), to = sample(1:10, 8),
      label = paste("interaction type", 1:8),
      length = c(100, 500),
      width = c(4, 1),
      arrows = c("to", "from", "middle", "middle;to"),
      dashes = c(TRUE, FALSE),
      title = paste("interaction name", 1:8),
      smooth = c(FALSE, TRUE),
      shadow = c(FALSE, TRUE, FALSE, TRUE)
    )
    
# define co-ordinates manually
    coords <- as.matrix(data.frame(
      x = c(-319,267,-641,321,-580,277,-418,166,-574,314),
      y = c(-168,169,-65,42,7,-175,88,205,-161,-82),
      stringsAsFactors = FALSE
    ))
    
    nodes <- data.frame(
      id = 1:10, group = c("A", "B"),
      label = paste("Node", 1:10),
      shape = "ellipse"
    )
    
    visNetwork::visNetwork(nodes, edges, height = "500px", width = "100%") %>%
  visNetwork::visIgraphLayout(layout = "layout.norm", layoutMatrix = coords)        
  })           
}

ui <- fluidPage(
  visNetwork::visNetworkOutput("network")      
)

shinyApp(ui = ui, server = server)

---

data

> nodes
   id group   label   shape
1   1     A  Node 1 ellipse
2   2     B  Node 2 ellipse
3   3     A  Node 3 ellipse
4   4     B  Node 4 ellipse
5   5     A  Node 5 ellipse
6   6     B  Node 6 ellipse
7   7     A  Node 7 ellipse
8   8     B  Node 8 ellipse
9   9     A  Node 9 ellipse
10 10     B Node 10 ellipse

> edges
  from to              label length width    arrows dashes              title smooth shadow
1    4  1 interaction type 1    100     4        to   TRUE interaction name 1  FALSE  FALSE
2    6 10 interaction type 2    500     1      from  FALSE interaction name 2   TRUE   TRUE
3    1  6 interaction type 3    100     4    middle   TRUE interaction name 3  FALSE  FALSE
4    8  3 interaction type 4    500     1 middle;to  FALSE interaction name 4   TRUE   TRUE
5   10  4 interaction type 5    100     4        to   TRUE interaction name 5  FALSE  FALSE
6    9  5 interaction type 6    500     1      from  FALSE interaction name 6   TRUE   TRUE
7    2  9 interaction type 7    100     4    middle   TRUE interaction name 7  FALSE  FALSE
8    3  8 interaction type 8    500     1 middle;to  FALSE interaction name 8   TRUE   TRUE

---

Current Output : In the above code the coords df with the x,y values are manually created. The values for x and y were chosen by trial and error by checking which values separate the nodes for respective groups. The 2 groups in the data are Blue nodes ( group A) and Yellow nodes ( group B). However, since I have large data set this manual approach is not possible. I am looking for a way to automate the assignment of x,y co-ordinates to each group programmatically and still get the groups separated. The desired output below is something I'm aiming towards with cell boundaries and then nodes arranged within the cell. Appreciate any inputs.

Desired Output

from gather nodes based on group information