usccana / netplot Goto Github PK

View Code? Open in Web Editor NEW

49.0 5.0 2.0 107.89 MB

Beautiful graph drawing

Home Page: https://usccana.github.io/netplot/

License: Other

R 98.02% Dockerfile 0.37% Makefile 1.61%

rstats rpackage network-analysis network-visualization igraph

netplot's Introduction

netplot

An alternative graph visualization tool that emphasizes aesthetics, providing default parameters that deliver out-of-the-box lovely visualizations.

Some features:

Auto-scaling of vertices using sizes relative to the plotting device.
Embedded edge color mixer.
True curved edges drawing.
User-defined edge curvature.
Nicer vertex frame color.
Better use of space-filling the plotting device.

The package uses the grid plotting system (just like ggplot2).

Installation

You can install the released version of netplot from CRAN with:

install.packages("netplot")

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("USCCANA/netplot")

Example

This is a basic example which shows you how to solve a common problem:

library(igraph)
#> 
#> Attaching package: 'igraph'
#> The following objects are masked from 'package:stats':
#> 
#>     decompose, spectrum
#> The following object is masked from 'package:base':
#> 
#>     union
library(netplot)
#> Loading required package: grid
#> 
#> Attaching package: 'netplot'
#> The following object is masked from 'package:igraph':
#> 
#>     ego
set.seed(1)
data("UKfaculty", package = "igraphdata")
l <- layout_with_fr(UKfaculty)
#> This graph was created by an old(er) igraph version.
#>   Call upgrade_graph() on it to use with the current igraph version
#>   For now we convert it on the fly...

plot(UKfaculty, layout = l) # ala igraph

V(UKfaculty)$ss <- runif(vcount(UKfaculty))
nplot(UKfaculty, layout = l) # ala netplot

sna::gplot(intergraph::asNetwork(UKfaculty), coord=l)

UKfaculty

# Random names
set.seed(1)
nam <- sample(babynames::babynames$name, vcount(UKfaculty))

ans <- nplot(
  UKfaculty,
  layout                = l,
  vertex.color          = ~ Group,
  vertex.label          = nam,
  vertex.size.range     = c(.01, .04, 4),
  vertex.label.col      =  "black",
  vertex.label.fontface = "bold",
  bg.col                = "transparent",
  vertex.label.show     = .5,
  vertex.label.range    = c(10, 25),
  edge.width.range      = c(1, 4, 5)
  )


# Plot it!
ans

Starting version 0.2-0, we can use gradients!

ans |>
  set_vertex_gpar(
    element = "core",
    fill = lapply(get_vertex_gpar(ans, "frame", "col")$col, \(i) {
      radialGradient(c("white", i), cx1=.8, cy1=.8, r1=0)
      }))

USairports

# Loading the data
data(USairports, package="igraphdata")

# Generating a layout naively
layout   <- V(USairports)$Position
#> This graph was created by an old(er) igraph version.
#>   Call upgrade_graph() on it to use with the current igraph version
#>   For now we convert it on the fly...
layout   <- do.call(rbind, lapply(layout, function(x) strsplit(x, " ")[[1]]))
layout[] <- stringr::str_remove(layout, "^[a-zA-Z]+")
layout   <- matrix(as.numeric(layout[]), ncol=2)

# Some missingness
layout[which(!complete.cases(layout)), ] <- apply(layout, 2, mean, na.rm=TRUE)

# Have to rotate it (it doesn't matter the origin)
layout <- netplot:::rotate(layout, c(0,0), pi/2)

# Simplifying the network
net <- simplify(USairports, edge.attr.comb = list(
  weight = "sum",
  name   = "concat",
  Passengers = "sum",
  "ignore"
))

# Pretty graph
nplot(
  net,
  layout            = layout,
  edge.width        = ~ Passengers,
  edge.color        = 
    ~ ego(col = "white", alpha = 0) + alter(col = "yellow", alpha = .75),
  skip.vertex       = TRUE,
  skip.arrows       = TRUE,
  edge.width.range  = c(.75, 4, 4), 
  bg.col            = "black",
  edge.line.breaks  = 10
  )

netplot's People

Stargazers

Watchers

Forkers

baifengbai anhnguyendepocen

netplot's Issues

Empty graph fails to plot

library(ergm)
library(netplot)

n <- 50
n_egos <- 10
net <- as.network(matrix(0, ncol = n, nrow = n), directed = TRUE)
nplot(net)
# Error in quantile.default(netenv$vertex.size, 1 - netenv$vertex.label.show[1]) : 
#  missing values and NaN's not allowed if 'na.rm' is FALSE

Implement a node no-overlapping algorithm

Using Spring Algorithms to Remove Node Overlapping (https://dl.acm.org/citation.cfm?id=1082334)

https://stackoverflow.com/questions/3265986/an-algorithm-to-space-out-overlapping-rectangles

remove arrows from edges for undirected graphs

Hi,

Thank you so much for building an incredible network visualization tool! I'm curious if I am missing something. I have an undirected network that I'd like to visualize without edge arrows. I have set skip.arrows=T, but the network plot still shows edge arrows. Is there a straightforward way to remove the arrows that I am missing?

I am sorry if I am overlooking something obvious.

Thank you so much.

Issue in making 'vertex.nsides' be determined by function

Hey @gvegayon! As discussed, here is the error we bumped into.

library(igraph)
#> 
#> Attaching package: 'igraph'
#> The following objects are masked from 'package:stats':
#> 
#>     decompose, spectrum
#> The following object is masked from 'package:base':
#> 
#>     union
library(netplot)
#> Loading required package: grid
#> 
#> Attaching package: 'netplot'
#> The following object is masked from 'package:igraph':
#> 
#>     ego

set.seed(123)
g <- erdos.renyi.game(50, 0.2)

teams <- c("Utah", "Wyoming", "Idaho", "Nevada")
V(g)$team <- sample(teams, vcount(g), replace=TRUE) 

# Map teams to sides
team_sides <- c(Utah=4, Wyoming=6, Idaho=3, Nevada=5)
sides <- team_sides[V(g)$team]

# Map teams to colors
team_colors <- c(Utah="red", Wyoming="darkgreen", Idaho="blue", Nevada="purple")

# Get color for each vertex 
vertex_colors <- team_colors[V(g)$team]

# Plot
# Create empty plot
plot(0,type="n", axes=FALSE)

nplot(g,
      vertex.color = ~team,  
      vertex.nsides= ~team)
#> Error in eval(netenv$vertex.nsides, envir = data): invalid 'envir' argument of type 'closure'

^{Created on 2023-10-04 with reprex v2.0.2}

Passing layout as other than matrix fails

Add option vertex.shape

Should be the following:

circle (15)
triangle (3 + rotation)
diamond (4)
square (4 + rotation)
line (1)
pentagon (5)
hexagon (6)

This will work as a shortcut to vertex.nbreaks.

Using formulae to set gpar attributes

Right now we do have a function to parse formulas for edge colors, but ideally, we could have a more comprehensive/flexible system to parse formulas and return various types of outcomes.

About the input

Formulas for mixing ego/alter attributes.
Formulas for using either ego or alter attributes.
Formulas using edge attributes.

Mixing these three should be relatively straightforward. Since at the edge level, we will only find one attribute of each.

The problem is when, for example, we would like to generate a statistic at the ego-network level.

About the output

Formulas should return numbers/factors:
- Integers (factors)
- Fractions
- Strings
- Etc.

This could follow a goal similar to what the scales R package does, this is, return palettes/functions for plots.

enable \n in vertex labels

Fantastic Package! Vertex labels with new lines are enabled in igraph plotting but not in this package. It would increase my use greatly if I could include new lines in my vertex labels like "Com 12\nMemmory T cells". Maybe a bug, maybe a enhancement... not sure.

p <- nplot(...)
p <- set_vertex_gpar(p, "core", fill = radialGradient(c("white", "black"), cx1=.8, cy1=.8, r1=0))
print(p) # fails