glow

A package for making glow-y plots

The glow package is a framework for creating plots with glowing points as an alternative way of plotting large point clouds.

Gallery

Methylation 450K Volcano Plot	Diamonds

Milky Way Galaxy (6.1 million stars)

OpenStreetMap GPS traces (2.8 billion points)

Clifford strange attractor (1 billion points)

Airline Dataset (145 million points)	Glow-y Spiral

U.S. Coronavirus Cases (2021)

Installation

remotes::install_github("traversc/glow")

Some advantages over traditional techniques

Naturally displays point density
glow plots don’t depend on the order of points in the data (points are commutative and associative)
Multi-threaded, can be faster than geom_point depending on settings
No loss of individual points compared to binning procedures
Naturally works with larger-than-memory datasets (See “Airline” dataset in inst/examples/examples.r)

Usage

Creating a glow plot is done through the GlowMapper or GlowMapper4 classes, which utilize the R6 class framework.

The class function $map creates a raster that can be plotted with ggplot’s geom_raster or output directly using the EBImage library.

See the help files and inst/examples/notes.txt for more information on each example.

ggplot example using the diamonds dataset

library(glow)
library(ggplot2)
library(viridisLite) # Magma color scale

# Number of threads
nt <- 4

data(diamonds)
gm <- GlowMapper$new(xdim=800, ydim = 640, blend_mode = "screen", nthreads=nt)

# relx(0.002) makes point size relative to x-axis, e.g. each point radius is 0.2% of the y-axis
gm$map(x=diamonds$carat, y=diamonds$price, intensity=1, radius = rely(0.002))
pd <- gm$output_dataframe(saturation = 1)

# Dark color theme
ggplot() + 
  geom_raster(data = pd, aes(x = pd$x, y = pd$y, fill = pd$value), show.legend = FALSE) +
  scale_fill_gradientn(colors = additive_alpha(magma(12))) +
  coord_fixed(gm$aspect(), xlim = gm$xlim(), ylim = gm$ylim()) + 
  labs(x = "carat", y = "price") + 
  theme_night(bgcolor = magma(12)[1])

# light "heat" color theme
light_colors <- light_heat_colors(144)
ggplot() + 
  geom_raster(data = pd, aes(x = pd$x, y = pd$y, fill = pd$value), show.legend = FALSE) +
  scale_fill_gradientn(colors = additive_alpha(light_colors)) +
  coord_fixed(gm$aspect(), xlim = gm$xlim(), ylim = gm$ylim()) + 
  labs(x = "carat", y = "price") + 
  theme_bw(base_size = 14)

# light "cool" color theme
light_colors <- light_cool_colors(144)
ggplot() + 
  geom_raster(data = pd, aes(x = pd$x, y = pd$y, fill = pd$value), show.legend = FALSE) +
  scale_fill_gradientn(colors = additive_alpha(light_colors)) +
  coord_fixed(gm$aspect(), xlim = gm$xlim(), ylim = gm$ylim()) + 
  labs(x = "carat", y = "price") + 
  theme_bw(base_size = 14)

Writing a raster image directly

Instead of using ggplot, you can also output a raster image directly using the EBImage Bioconductor library.

library(EBImage)

# Generate data
cliff_points <- clifford_attractor(1e6, 1.886,-2.357,-0.328, 0.918, 0.1, 0)
color_pal <- circular_palette(n=144, pal_function=rainbow)
cliff_points$color <- map_colors(color_pal, cliff_points$angle, min_limit=-pi, max_limit=pi)

# Create raster
gm <- GlowMapper4$new(xdim=480, ydim = 270, blend_mode = "additive", nthreads=4)
gm$map(x=cliff_points$x, y=cliff_points$y, radius=1e-3, color=cliff_points$color)
pd <- gm$output_raw(saturation = 1)

# Output raster with EBImage
image_array <- array(1, dim=c(480, 270, 3))
image_array[,,1] <- pd[[1]]*pd[[4]]
image_array[,,2] <- pd[[2]]*pd[[4]]
image_array[,,3] <- pd[[3]]*pd[[4]]
img <- EBImage::Image(image_array, colormode='Color')
plot(img)
writeImage(img, "plots/clifford_vignette.png")