In martinjhnhadley/oidnChaRts: Utility Functions For htmlwidget Visualisations

Overview of function

scatter_plot can be used to create the following kinds of charts:

Scatter plots

Note that this function specifically assumes a high point density where marker opacity is required. If you're interested in comparing multiple series of points together with different shapes, you will need to use scatter_chart instead.

As with all oidnChaRts libraries, you are advised to load the htmlwidget library you're using directly.

data_stacked_bar_chart

library(oidnChaRts)

This vignette covers the use of scatter plots for visualising data with a variety of htmlwidget libraries, for demonstration purposes we use the following dataset generated from https://doi.org/10.6084/m9.figshare.4555441. To explain the details of the research dataset we would require domain specific expertise, but the format of the dataset in this package is fairly easy to understand:

head(data_scatter_plot)

The columns may be summarised as follows:

chromsome: Chromosome and genomic position of gene
type: Concordant (both sites are identical) or Discordant (mutations only in one replicate)
x: Allelic frequencies of gene in replicate 1
y: Allelic frequencies of gene in replicate 2

Crucial to this dataset is the comparison of the two traces, "Concordant" and "Discordant". This requires the scatter plot to use an alpha channel (i.e. transparency) to show the overlap/similarity of the two traces, as shown in the example highcharter visualisation below:

library(highcharter)
data_scatter_plot %>% mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, max = 255), rgb(203, 24, 29, max = 255))
  )) %>%
  scatter_plot(library = "highcharter",
               x.column = ~x,
               y.column = ~y,
               color.column = ~color,
               traces.column = ~type) %>%
  hc_xAxis(title = list(text = "Local AF (R1)"),
               gridLineWidth = 1,
               min = 0,
               max = 1,
               tickInterval = 0.25) %>%
  hc_yAxis(min = 0,
               max = 1,
               tickInterval = 0.25,
               title = list(text = "Local AF (R2)"))

Scatter plot specifications

In scatter plots it is important to consider the following data:

traces: distinct groups of points, requiring distinct legend items.
colors: color attributes of the markers and legend items.

The oidnChaRts scatter_plot function requires both a trace.column and color.column to be specified, because of oddities in many of the htmlwidget libraries in how colours are assigned at the marker and trace level. We add this information to the dataset using the tidyverse as follows:

data_scatter_plot %>% mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, max = 255), rgb(203, 24, 29, max = 255))
  ))

Scatter plot

The following will create a generic scatter plot with highcharter, note that internally oidnChaRts is using the Bubble Chart highchart type to allow point transparency.

library(highcharter)
data_scatter_plot %>% mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, max = 255), rgb(203, 24, 29, max = 255))
  )) %>%
  scatter_plot(library = "highcharter",
               x.column = ~x,
               y.column = ~y,
               color.column = ~color,
               traces.column = ~type)

Highcharter is slightly slow at displaying ~2000 bubbles, to improve the feel of the chart we could disable animation:

data_scatter_plot %>% mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, 0.8, max = 255), rgb(203, 24, 29, 0.8, max = 255))
  )) %>%
  scatter_plot(library = "highcharter",
               x.column = ~x,
               y.column = ~y,
               color.column = ~color,
               traces.column = ~type) %>%
  hc_plotOptions(series = list(animation = FALSE))

The following will create a generic scatter plot with plotly, which is significantly faster than highcharter:

library(plotly)
data_scatter_plot %>% mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, 0.5, max = 255), rgb(203, 24, 29, 0.5, max = 255))
  )) %>%
  scatter_plot(library = "plotly",
               x.column = ~x,
               y.column = ~y,
               color.column = ~color,
               traces.column = ~type)

Tooltips

The highcharter library embeds the entirety of the data.frame into the scatter plot, and allows us to construct a tooltip as shown below.

library(dplyr)
data_scatter_plot %>%
  mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, 1, max = 255), rgb(203, 24, 29, 1, max = 255))
  )) %>%
scatter_plot(library = "highcharter",
           x.column = ~x,
           marker.size = 10,
           y.column = ~y,
           traces.column = ~type,
           color.column = ~color) %>%
  hc_tooltip(
        formatter = JS(
          "function(){
          return '<b>' + this.series.name + '</b>' + 
                  '<br/>' +
                  '<b>Chromosomal Position: </b>' + this.point.chromosome +
                  '<br/>' +
                  '<b>x: </b>' + this.x +
                  '<br/>' +
                  '<b>y: </b>' + this.y;}"
          )
        )

Unfortunately, the plotly library requires tooltips to be specified in the original plot_ly call, as th oidnChaRt library is primarily for demonstration purposes there are no plans to specially support plotly tooltips. Instead, consider creating a plotly chart from scratch as follows:

data_scatter_plot %>%
  mutate(color = plyr::mapvalues(
    type,
    from = c("Discordant", "Concordant"),
    to = c(rgb(251, 106, 74, 1, max = 255), rgb(203, 24, 29, 1, max = 255))
  )) %>%
  plot_ly(
    data = .,
    x = ~x,
    y = ~y,
    color = ~type,
    colors = unique(.[["color"]]),
    text = ~paste("Chromosomal position: ", chromosome)
  ) %>%
  add_markers(alpha = 0.5)