Highlighting with click/hover in gl2d seems glitchy

[myles-lewis]

I know this is supposed to have been fixed in plotly.js but I find that the 2d scatter plot using either type='scattergl' or toWebGL() breaks the plotly_click event in plotly.r. It works fine for me for scatter3d. I want to show a 2d plot of 21,000 points, hence the need for scattergl, and link to 2ndary boxplot.

This issue was reported here: plotly/plotly.js#743
but is supposed to have been completed here: plotly/plotly.js#994

Here is demo code - the lower window updates the event information in scatter3d but not for 2d. The boxplots link in 3d but not 2d:

# Dummy data
  pathotype.z <- data.frame(x=rnorm(15000), y=rnorm(15000), logP=rnorm(15000)^2, hcol=sample(1:5, 15000, replace=T), 
                          gene=paste0("A", as.integer(rnorm(15000)*10000)))

genelookup <- list()
  genelookup <- tapply(pathotype.z$gene, pathotype.z$hcol, function(x) x)

  ui <- fluidPage(fluidRow(column(8, plotlyOutput("volc3d", height=640)), 
                         column(4, plotlyOutput("scatterplot", height=500),
                                fluidRow(
                                  column(12, radioButtons("plottype", label = h5("Plot"),
                                                         choices = list("2d" = 1, "3d" = 2), selected = 2))
                                  ))),
        verbatimTextOutput("selection"))

  server <- function(input, output, session) {
  output$volc3d <- renderPlotly({
    pt <- as.numeric(input$plottype)
    if (length(pt)==0) pt <- 2
    if (pt==2) {
      plot_ly(d=pathotype.z, x=~x, y=~y, z=~logP, 
              marker=list(size=2.5), color=~hcol, 
              text=~gene, type='scatter3d', mode = 'markers', source='volc3d')
    } else {
      plot_ly(data=pathotype.z, x=~x, y=~y, 
                  marker=list(size=5), color=~hcol,
                  text=~gene, type="scattergl", mode = "markers")
    }})
  
  output$selection <- renderPrint({
    s <- event_data("plotly_click", source = "volc3d")
    if (length(s) == 0) {"Click on gene to display scatterplot"
    } else {
    as.list(s)
    }})
  
  output$scatterplot <- renderPlotly({
    s <- event_data("plotly_click", source = "volc3d")
    if (length(s)) {
      p <- s[["pointNumber"]]
      c <- s[["curveNumber"]]
      gene <- genelookup[[c+1]][p+1]
      plot_ly(pathotype.z, x=~hcol, y=~y, type='box') %>%
        layout(title=gene, yaxis=list(title=gene), 
               margin=list(t=100,l=50,r=50,b=50))
    } else {plotly_empty()}
  })
  }

  shinyApp(ui, server)

Here is session info:

> sessionInfo()
R version 3.3.1 (2016-06-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:
[1] LC_COLLATE=English_United Kingdom.1252  LC_CTYPE=English_United Kingdom.1252    LC_MONETARY=English_United Kingdom.1252
[4] LC_NUMERIC=C                            LC_TIME=English_United Kingdom.1252    

attached base packages:
 [1] grid      parallel  stats4    stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] gplots_3.0.1               ComplexHeatmap_1.10.2      ggforce_0.1.1              shiny_0.14.2              
 [5] plotly_4.5.6               ggplot2_2.2.0              DESeq2_1.12.4              SummarizedExperiment_1.2.3
 [9] Biobase_2.32.0             GenomicRanges_1.24.3       GenomeInfoDb_1.8.7         IRanges_2.6.1             
[13] S4Vectors_0.10.3           BiocGenerics_0.18.0       

loaded via a namespace (and not attached):
 [1] bitops_1.0-6         RColorBrewer_1.1-2   httr_1.2.1           prabclus_2.2-6       tools_3.3.1          R6_2.2.0            
 [7] KernSmooth_2.23-15   rpart_4.1-10         Hmisc_4.0-1          DBI_0.5-1            lazyeval_0.2.0       colorspace_1.3-2    
[13] trimcluster_0.1-2    nnet_7.3-12          GetoptLong_0.1.5     gridExtra_2.2.1      base64_2.0           htmlTable_1.7       
[19] labeling_0.3         caTools_1.17.1       diptest_0.75-7       scales_0.4.1         DEoptimR_1.0-8       mvtnorm_1.0-5       
[25] robustbase_0.92-7    genefilter_1.54.2    stringr_1.1.0        digest_0.6.10        foreign_0.8-67       XVector_0.12.1      
[31] base64enc_0.1-3      htmltools_0.3.5      htmlwidgets_0.8      GlobalOptions_0.0.10 RSQLite_1.1-1        shape_1.4.2         
[37] jsonlite_1.1         mclust_5.2           BiocParallel_1.6.6   gtools_3.5.0         acepack_1.4.1        dendextend_1.3.0    
[43] dplyr_0.5.0          RCurl_1.95-4.8       magrittr_1.5         modeltools_0.2-21    Formula_1.2-1        Matrix_1.2-7.1      
[49] Rcpp_0.12.8          munsell_0.4.3        yaml_2.1.14          stringi_1.1.2        whisker_0.3-2        MASS_7.3-45         
[55] zlibbioc_1.18.0      flexmix_2.3-13       plyr_1.8.4           gdata_2.17.0         udunits2_0.13        lattice_0.20-33     
[61] splines_3.3.1        annotate_1.50.1      circlize_0.3.9       locfit_1.5-9.1       knitr_1.15.1         rjson_0.2.15        
[67] fpc_2.1-10           geneplotter_1.50.0   XML_3.98-1.5         latticeExtra_0.6-28  data.table_1.10.0    tweenr_0.1.5        
[73] httpuv_1.3.3         gtable_0.2.0         openssl_0.9.5        purrr_0.2.2          tidyr_0.6.0          kernlab_0.9-25      
[79] assertthat_0.1       mime_0.5             xtable_1.8-2         class_7.3-14         survival_2.39-4      viridisLite_0.1.3   
[85] tibble_1.2           AnnotationDbi_1.34.4 memoise_1.0.0        units_0.4-1          cluster_2.0.5   

[cpsievert]

Hmm, these events seem to fire, but the behavior seems a bit wonky when used with highlight():

library(plotly)
library(crosstalk)

m <- SharedData$new(mtcars)

plot_ly(m, x = ~wt, y = ~mpg) %>%
  toWebGL() %>%
  highlight("plotly_hover")

Interestingly, the event data seems to be properly formatted -- http://codepen.io/cpsievert/pen/vxggOv

[freestatman]

Wanted to point out that, there is a bug in @Uncle-M 's example, we should add source= argument for 2D plot:
text=~gene, type="scattergl", mode = "markers", source='volc3d')

However, it seems that the clicked gene id doesn't match with the boxplot gene id, for both 2D and 3D plot. There maybe a mismatch in indexes. Wondering how s[["pointNumber"]] and s[["curveNumber"]] was defined within plotly?

[cpsievert]

This appears to be fixed now