Naming conventions

One of the deliberate choices in splot() is a consistent, predictable parameter naming system. Every parameter is prefixed by the element it controls – node_, edge_, label_, donut_, pie_, edge_label_, edge_ci_ – followed by the property: size, color, fill, width, alpha, and so on. So node_fill is the node fill color, edge_label_size is the edge label font size, donut_border_color is the donut border color, and edge_ci_alpha is the confidence interval underlay transparency. The naming is designed so that once you learn the pattern, you can guess parameter names without checking the documentation. This contrasts with conventions like vsize, posCol, border.width, label.cex, and pieColor, where every parameter follows a different convention and needs to be memorized individually.

Statistical annotations

Network plots in research often need to communicate uncertainty and significance alongside the graph structure itself. splot() treats this as a first-class concern rather than an afterthought. Edges can carry confidence interval underlays (edge_ci) – semi-transparent bands behind each edge whose width reflects the uncertainty of the estimate. Edge labels can be built from templates (edge_label_template) using placeholders like {est}, {stars}, {low}, {up}, {range}, and {p}, which are filled in from vectors of CI bounds and p-values that you supply. Significance stars are computed automatically from p-values when edge_label_stars = TRUE. These features mean that a single splot() call can produce a network where every edge shows its weight, its significance level, and its confidence interval – information that would otherwise require post-processing or manual annotation.

Inputs and compatibility

splot() accepts square adjacency matrices, data frames with edge lists (from, to, weight columns), igraph objects, cograph network objects, tna objects, and qgraph objects. For users migrating from qgraph, the from_qgraph() function converts a fitted qgraph object into splot parameters, and tplot() provides a wrapper that accepts qgraph-style parameter names (vsize, edge.labels, posCol) for a smoother transition.

library(cograph)
library(qgraph)

Node parameters

Edge parameters

Layout and display

Pie and donut

CI underlays

net <- cograph(mat)
ne <- nrow(get_edges(net))

set.seed(42)
ci_widths <- runif(ne, 0.1, 0.4)

splot(mat, node_size = 9,
      edge_ci = ci_widths,
      edge_ci_scale = 5,
      edge_ci_alpha = 0.3)

CI range labels

ci_lower <- runif(ne, 0.01, 0.10)
ci_upper <- runif(ne, 0.20, 0.50)

splot(mat, node_size = 9,
      edge_label_template = "{est} [{low}, {up}]",
      edge_ci_lower = ci_lower,
      edge_ci_upper = ci_upper,
      edge_label_size = 0.4)

Edge label templates with significance stars

p_values <- round(runif(ne, 0.0001, 0.08), 4)

splot(mat, node_size = 9,
      edge_label_template = "{est}{stars}",
      edge_label_p = p_values,
      edge_label_stars = TRUE)

Pie chart nodes

set.seed(42)
pie_vals <- lapply(1:9, function(i) {
  v <- runif(3)
  v / sum(v)
})

splot(mat, node_size = 9,
      pie_values = pie_vals,
      pie_colors = rep(list(c("#E41A1C", "#377EB8", "#4DAF4A")), 9))

Polygon donut shapes

splot(mat, node_size = 9,
      donut_fill = fills,
      donut_color = palette_rainbow(9),
      donut_shape = "hexagon")

Double donuts

splot(mat, node_size = 9,
      donut_fill = fills,
      donut_color = palette_rainbow(9),
      donut2_values = lapply(1:9, function(i) runif(1)),
      donut2_colors = lapply(palette_colorblind(9), function(x) x))