Skip to contents

Network Visualization with cograph: A Complete Plotting Guide

2026-03-05

Source: vignettes/articles/1_cograph-tutorial-plotting.qmd

1 Introduction

Network visualization in R has traditionally meant choosing between two extremes: quick-and-dirty plots that look like hairballs, or heavily customized code that takes longer to write than the analysis itself. cograph collapses this tradeoff. The splot() function produces publication-ready network visualizations from a single call, with sensible defaults that you can override parameter by parameter.

This tutorial demonstrates everything through a single TNA model of collaborative regulation behaviors. We start with a zero-argument plot and progressively layer in customization — so you can see exactly what each parameter changes.

  1. The plot — one line, zero arguments, publication-ready
  2. Node styling — shapes, colors, sizes, labels, centrality-based scaling
  3. Edge styling — width, color, curvature, labels, thresholds
  4. Donuts and pies — composite node shapes for multivariate data
  5. Themes and palettes — pre-built visual themes
  6. Pipe-friendly API — the sn_* function family for chainable plot building
  7. Specialized plots — chord diagrams, heatmaps, alluvial flows
  8. Group comparisons — difference networks and multi-group grids
  9. Bootstrap — validated network visualization
  10. Permutation testing — significant group differences
  11. Mixed networks — directed and undirected edges combined

For TNA-specific workflows, see the TNA tutorial.

GitHub Development Version

This tutorial covers the full feature set available in the GitHub development version of cograph. The CRAN release provides the core plotting engine; the development version adds themes, pipe API, specialized plots, and additional features.

# Development version with full features
# devtools::install_github("sonsoleslp/cograph")
# CRAN version
install.packages("cograph")

2 Building the Model

We use the built-in group_regulation dataset from the tna package — coded collaborative regulation behaviors from student groups. This gives us a 9-state directed transition network, the kind of rich structure that showcases cograph’s capabilities.

library(tna)
data("group_regulation")
model <- tna(group_regulation)
model
State Labels :

   adapt, cohesion, consensus, coregulate, discuss, emotion, monitor, plan, synthesis

Transition Probability Matrix :

                  adapt   cohesion  consensus coregulate    discuss    emotion
adapt      0.0000000000 0.27308448 0.47740668 0.02161100 0.05893910 0.11984283
cohesion   0.0029498525 0.02713864 0.49793510 0.11917404 0.05958702 0.11563422
consensus  0.0047400853 0.01485227 0.08200348 0.18770738 0.18802338 0.07268131
coregulate 0.0162436548 0.03604061 0.13451777 0.02335025 0.27360406 0.17208122
discuss    0.0713743356 0.04758289 0.32118451 0.08428246 0.19488737 0.10579600
emotion    0.0024673951 0.32534367 0.32040888 0.03419105 0.10186817 0.07684173
monitor    0.0111653873 0.05582694 0.15910677 0.05792045 0.37543615 0.09071877
plan       0.0009745006 0.02517460 0.29040117 0.01721618 0.06789021 0.14682475
synthesis  0.2346625767 0.03374233 0.46625767 0.04447853 0.06288344 0.07055215
              monitor       plan   synthesis
adapt      0.03339882 0.01571709 0.000000000
cohesion   0.03303835 0.14100295 0.003539823
consensus  0.04661084 0.39579712 0.007584137
coregulate 0.08629442 0.23908629 0.018781726
discuss    0.02227284 0.01164262 0.140976968
emotion    0.03630596 0.09975326 0.002819880
monitor    0.01814375 0.21563154 0.016050244
plan       0.07552379 0.37420822 0.001786584
synthesis  0.01226994 0.07515337 0.000000000

Initial Probabilities :

     adapt   cohesion  consensus coregulate    discuss    emotion    monitor
    0.0115     0.0605     0.2140     0.0190     0.1755     0.1515     0.1440
      plan  synthesis
    0.2045     0.0195

We also prepare a group model for later comparison and permutation sections:

data("group_regulation_long")
prepared <- prepare_data(group_regulation_long,
                         action = "Action", actor = "Actor", time = "Time")
group_model <- group_tna(prepared, group = "Achiever")

3 The Plot

One function. Zero arguments beyond the model. Publication-ready.

splot(model)
Figure 1: splot(model) — that’s it

When cograph receives a TNA object, it auto-applies TNA-appropriate defaults: oval layout, dark blue edges weighted by transition probability, donut rings showing initial state probabilities, and directional arrows. Every aspect of this plot can be customized — the rest of this tutorial shows how.

4 Node Styling

4.1 Shapes

Every node can have its own shape. cograph supports circles, squares, hexagons, diamonds, triangles, pentagons, stars, hearts, and more:

splot(model,
  node_shape = c("circle", "square", "hexagon", "diamond", "triangle",
                 "pentagon", "star", "heart", "circle"),
  layout = "circle"
)
Figure 2: Different shapes per node

All Built-In Shapes

"circle", "square", "triangle", "diamond", "pentagon", "hexagon", "star", "heart", "ellipse", "cross", "rectangle", "pie", "donut"

You can also register custom SVG shapes with register_svg_shape() and use them by name.

4.2 Colors and Sizes 

splot(model,
  node_fill = palette_viridis(9),
  node_border_color = "white",
  node_border_width = 2,
  node_size = c(12, 8, 10, 9, 7, 11, 8, 10, 9)
)
Figure 3: Custom fills, borders, and per-node sizes

4.3 Scaling Nodes by Centrality

scale_nodes_by maps a centrality measure to node size — larger nodes are more central. This is one of the most useful parameters for making network structure immediately visible:

par(mfrow = c(1, 2), mar = c(1, 1, 2, 1))
splot(model, scale_nodes_by = "betweenness", node_size_range = c(3, 14),
      title = "Betweenness")
splot(model, scale_nodes_by = "strength", node_size_range = c(3, 14),
      title = "Strength")
Figure 4: Node size proportional to different centrality measures

scale_nodes_by Options

Any centrality measure name works: "degree", "strength", "betweenness", "closeness", "eigenvector", "pagerank", "authority", "hub", "eccentricity", "coreness", "constraint", "transitivity", "harmonic", "diffusion", "leverage", "kreach".

For directional measures (in/out), use a list: scale_nodes_by = list("strength", mode = "in").

Parameter Default Description
scale_nodes_by NULL Centrality measure name or list with mode
node_size_range c(2, 8) c(min, max) size range for mapping

5 Edge Styling

5.1 Width and Curvature 

splot(model,
  edge_width = 3,
  curvature = 0.3,
  arrow_size = 0.015
)
Figure 5: Custom edge width range with curvature

5.2 Edge Labels

Display transition probabilities directly on edges:

splot(model,
  edge_labels = TRUE,
  edge_label_size = 0.6,
  edge_label_bg = "white",
  threshold = 0.05
)
Figure 6: Edge labels showing transition weights

5.3 Thresholding

Hide weak transitions to reveal the backbone:

par(mfrow = c(1, 2), mar = c(1, 1, 2, 1))
splot(model, title = "All edges")
splot(model, threshold = 0.10, title = "threshold = 0.10")
Figure 7: Thresholding removes visual clutter

Edge Parameters

Parameter Default Description
edge_width auto Base edge width
edge_width_range auto c(min, max) width range
edge_color auto Edge color
edge_alpha 1 Transparency
edge_style "solid" "solid", "dashed", "dotted"
curvature auto Curve amount
arrow_size auto Arrowhead size
threshold 0 Hide edges below this weight
edge_labels FALSE Show weight labels
edge_label_style "plain" "plain", "ci", "stars", "full"
edge_positive_color auto Color for positive weights
edge_negative_color auto Color for negative weights

6 Donuts and Pies

Donut and pie node shapes encode additional quantitative information directly in the node rendering. In TNA, the donut ring is automatically filled from initial state probabilities — but you can use it for any node-level metric.

6.1 Donut Nodes

The default TNA plot already shows donut fills from model$inits. You can customize the appearance:

splot(model, 
      donut_color = "black",
      donut_bg_color = "white",
      donut_border_color = "black",
      donut_shape = "circle"
)
Figure 8: Donuts with custom styling

6.2 Pie Nodes

Pie chart nodes show a categorical distribution per node. You can combine them with the outer donut ring:

# Each node gets a 3-slice pie (e.g., from a clustering or condition split)
pie_vals <- list(
  c(0.5, 0.3, 0.2), c(0.4, 0.4, 0.2), c(0.3, 0.3, 0.4),
  c(0.6, 0.2, 0.2), c(0.2, 0.5, 0.3), c(0.4, 0.3, 0.3),
  c(0.7, 0.2, 0.1), c(0.3, 0.4, 0.3), c(0.5, 0.2, 0.3)
)

splot(model,
  pie_values = pie_vals,
  pie_colors = c("#1976D2", "#FFA000", "#C62828"),
  layout = "circle"
)
Figure 9: Donut ring (initial probability) + inner pie segments

Donut & Pie Parameters

Parameter Default Description
donut_fill auto (from TNA inits) Numeric 0–1 per node
donut_color auto Fill color for the ring
donut_inner_ratio 0.5 Inner radius ratio (0–1)
donut_shape "circle" "circle", "square", "hexagon", "diamond", "triangle", "pentagon"
donut_show_value FALSE Show numeric value in center
donut_value_suffix "" Text after value (e.g., "%")
pie_values NULL List of numeric vectors per node
pie_colors auto Colors for pie segments

7 Themes and Palettes

Themes apply a coordinated set of visual defaults with a single parameter:

par(mfrow = c(2, 2), mar = c(1, 1, 2, 1))
splot(model, theme = "classic",    title = "classic")
splot(model, theme = "dark",       title = "dark")
splot(model, theme = "minimal",    title = "minimal")
splot(model, theme = "colorblind", title = "colorblind")
Figure 10: Four built-in themes 
par(mfrow = c(1, 3), mar = c(1, 1, 2, 1))
splot(model, node_fill = palette_viridis(9),    title = "viridis")
splot(model, node_fill = palette_pastel(9),     title = "pastel")
splot(model, node_fill = palette_colorblind(9), title = "colorblind")
Figure 11: Color palettes for nodes

All Themes

Theme Description
"classic" Clean, professional (default)
"dark" Dark background, light elements
"minimal" Reduced chrome, subtle colors
"colorblind" Colorblind-safe palette
"grayscale" No color, grayscale only
"vibrant" Bold, saturated colors
"nature" Earth-tone palette
"viridis" Viridis color scale

Use list_themes() to see all registered themes. Create custom themes with register_theme().

8 Pipe-Friendly API (ggplot2)

The sn_* function family lets you build plots incrementally using the pipe operator. Each function takes a cograph_network and returns a modified one:

model |>
  cograph(layout = "circle") |>
  sn_theme("dark") |>
  sn_nodes(size = 10, shape = "hexagon",
           border_color = "white", border_width = 2) |>
  sn_edges(width = 4, curvature = 0.2) |>
  soplot(title = "Built with pipes")
Figure 12: Dark-themed hexagon network built with pipes

All sn_* Functions

Function What It Configures
sn_layout(network, layout, seed) Layout algorithm
sn_theme(network, theme) Visual theme
sn_palette(network, palette, target, by) Color palette for nodes or edges
sn_nodes(network, ...) All node aesthetics (size, shape, fill, labels, pie, donut)
sn_edges(network, ...) All edge aesthetics (width, color, labels, curvature, CI)
sn_render(network, ...) Render with grid graphics (alias for soplot())
sn_ggplot(network, title) Convert to ggplot2 object
sn_save(network, filename, ...) Save to file (PDF/PNG/SVG)

9 Specialized Plots

9.1 Chord Diagram

Chord diagrams show flows as curved ribbons around a circle. The width of each ribbon is proportional to the transition weight:

plot_chord(model, chord_alpha = 0.6, title = "Transition Flows")
Figure 13: Chord diagram of transition flows

Chord Diagram Parameters

Parameter Default Description
segment_colors auto Colors for each node’s arc
chord_color_by "source" Color ribbons by "source", "target", or "weight"
chord_alpha 0.5 Ribbon transparency
self_loop FALSE Show self-transition arcs?
threshold 0 Hide chords below this weight

9.2 Heatmap

The adjacency matrix as a color-coded grid, optionally grouped by cluster:

clusters <- list(
  Social    = c("discuss", "consensus", "cohesion", "emotion"),
  Cognitive = c("plan", "monitor", "adapt", "synthesis"),
  Meta      = c("coregulate")
)

plot_heatmap(model,
  cluster_list = clusters,
  show_values = TRUE,
  value_size = 3.5,
  title = "Transition Heatmap"
)
Figure 14: Transition heatmap with clustered groups

9.3 Alluvial / Sankey Diagram

Shows source-to-target flow as an alluvial diagram:

plot_alluvial(model$weights, threshold = 0.05)
Figure 15: Alluvial diagram of transition flows

10 Group Comparisons

10.1 Multi-Group Grid

When you pass a group_tna object, splot renders all groups in a grid:

splot(group_model, i = NULL)
Figure 16: High vs Low achiever networks side by side

10.2 Difference Network

plot_compare() computes and visualizes the element-wise difference between two groups. Green edges are stronger in the first group; red edges are stronger in the second:

plot_compare(group_model,
  pos_color = "#2A9D8F",
  neg_color = "#E76F51",
  title = "High vs Low"
)
Figure 17: Difference network: High - Low achievers

11 Bootstrap

The bootstrap procedure determines which edges are stable properties of the data and which are sampling artifacts. Pass a tna_bootstrap object to splot() and it automatically styles significant vs. non-significant edges:

set.seed(265)
boot <- bootstrap(model, iter = 1000, level = 0.05)
splot(boot)
Figure 18: Bootstrap-validated network (solid = significant, dashed = non-significant) 
splot(boot, display = "significant")
Figure 19: Only significant edges shown 
splot(boot, display = "ci", threshold = 0.05)
Figure 20: Confidence interval labels on edges

Bootstrap Display Modes

display Description
"styled" (default) Significant = solid, non-significant = dashed
"significant" Only significant edges shown
"full" All edges shown with equal styling
"ci" Confidence interval labels on edges

Additional parameters: show_stars (significance stars), show_ci (CI values), width_by (scale width by weight or CI width).

12 Permutation Testing

Permutation testing evaluates whether between-group differences are statistically significant. Pass the result to splot() to visualize which edge differences survived the permutation test:

set.seed(265)
perm <- permutation_test(group_model, iter = 1000)
splot(perm)
Figure 21: Permutation test: significant between-group edge differences

Permutation Plot Parameters

Parameter Default Description
show_nonsig TRUE Show non-significant edges?
edge_nonsig_color "gray80" Color for non-significant edges
edge_nonsig_alpha 0.3 Transparency for non-significant edges
show_stars FALSE Show significance stars
show_effect FALSE Show effect sizes

13 Mixed Networks

Overlay directed and undirected edges on the same plot. This is useful when you have two types of relationships — for example, symmetric co-occurrence patterns and asymmetric transition probabilities. The defaults follow TNA conventions: oval layout, weight-scaled edges, dotted edge starts, and visible edge labels.

# Undirected: symmetric co-occurrence (half the weight for visual balance)
set.seed(42)
sym <- (model$weights + t(model$weights)) / 2
noise <- matrix(runif(length(sym), -0.02, 0.02), nrow = nrow(sym))
noise <- (noise + t(noise)) / 2  # keep symmetric
diag(noise) <- 0
sym <- pmax(sym * 0.5 + noise, 0)

# Directed: original transition probabilities
asym <- model$weights

# Defaults: oval layout, blue edges, weight-scaled widths, edge labels
plot_mixed_network(sym, asym, threshold = 0.03,
  title = "Co-occurrence + Transitions"
)
Figure 22: Co-occurrence (steel blue, straight) and transitions (dark blue, curved) with TNA-style defaults

You can distinguish edge types by color: undirected co-occurrence edges are straight steel-blue lines, while directed transitions are curved dark-blue arrows with dotted starts. Edge widths scale automatically by weight, so stronger connections stand out.

plot_mixed_network(sym, asym, threshold = 0.03,
  sym_color = "#E63946",
  asym_color = "#003355",
  title = "Red undirected + Blue directed"
)
Figure 23: Custom colors to visually separate edge types

14 Multi-Cluster Visualization

plot_mcml() draws two layers simultaneously: individual nodes inside cluster shells (bottom) and aggregated cluster-level summary pies (top).

clusters <- list(
  Social    = c("discuss", "consensus", "cohesion", "emotion"),
  Cognitive = c("plan", "monitor", "adapt", "synthesis"),
  Meta      = c("coregulate")
)

mat <- model$weights

plot_mcml(mat, clusters, mode = "tna",
  title = "Collaborative Regulation Clusters")
Figure 24: Two-layer hierarchical cluster view

plot_mtna() provides a flat single-layer alternative:

plot_mtna(mat, clusters, show_labels = TRUE, label_abbrev = 4,
  title = "Flat Cluster Map")
Figure 25: Flat cluster layout

15 Nestimate Integration

The Nestimate package estimates networks from sequence data using multiple methods. Its objects dispatch through splot() automatically.

library(Nestimate)

net <- build_network(group_regulation, method = "relative")
splot(net, title = "Nestimate: relative transitions")

Bootstrap stability:

set.seed(42)
nboot <- bootstrap_network(net, iter = 500)
splot(nboot, title = "Nestimate bootstrap")

Group comparison:

grp <- build_network(group_regulation_long, method = "relative",
  action = "Action", actor = "Actor", time = "Time",
  group = "Achiever")
splot(grp)

16 Motifs

Motif analysis identifies recurring local connectivity patterns — the building blocks of network structure. cograph uses the MAN (Mutual, Asymmetric, Null) classification for directed triads: 16 possible patterns from empty (003) to fully connected (300).

16.1 Census

motifs() counts how often each MAN type appears and tests significance against a configuration model null:

mot <- motifs(model, significance = TRUE, n_perm = 100, seed = 42)
mot
Motif Census
Level: individual | 2000 units | States: 9 | Pattern: triangle
Significance: permutation (n_perm=100)

Type distribution:

030C 030T 120C 120D 120U  210  300
   1    1    1    1    1    1    1

Top 7 results:
 type count expected     z      p   sig
 030T   620    435.4  8.14 0.0000  TRUE
  300    79    169.4 -7.44 0.0000  TRUE
  210   581    766.4 -6.52 0.0000  TRUE
 120C  1482   1356.5  3.40 0.0007  TRUE
 120U   190    169.7  1.55 0.1211 FALSE
 120D   178    172.9  0.46 0.6455 FALSE
 030C  1054   1063.6 -0.29 0.7718 FALSE

16.2 Visualization 

plot(mot, type = "significance")
Figure 26: Significance: z-scores against configuration model 
plot(mot, type = "patterns")
Figure 27: MAN triad pattern diagrams

16.3 Named subgraphs

subgraphs() identifies the specific node triples forming each pattern — which states participate in which motifs:

sg <- subgraphs(model, significance = TRUE, top = 10, pattern="triangle")
sg
Motif Subgraphs
Level: individual | 2000 units | States: 9 | Pattern: triangle
Significance: permutation (n_perm=1000)
Min count: > 5

Type distribution:

120C 030C 030T  210
   6    2    1    1

Top 10 results:
                         triad observed type expected       z p  sig
    consensus - discuss - plan      278  210   1504.1 -148.31 0 TRUE
   cohesion - consensus - plan      151 120C   1387.4 -109.94 0 TRUE
    consensus - emotion - plan      436 120C   1443.3 -105.12 0 TRUE
      discuss - emotion - plan       76 030T   1317.2 -103.78 0 TRUE
 consensus - coregulate - plan      301 120C   1395.1 -100.86 0 TRUE
    consensus - monitor - plan      187 120C   1336.8  -99.64 0 TRUE
  consensus - plan - synthesis       13 120C   1275.7  -99.07 0 TRUE
 consensus - discuss - emotion      238 120C   1346.5  -96.38 0 TRUE
     cohesion - discuss - plan       18 030C   1199.3  -88.93 0 TRUE
   coregulate - discuss - plan       61 030C   1217.0  -88.68 0 TRUE
plot(sg, n = 10, ncol = 5)
Figure 28: Top triads as network diagrams

17 Higher-Order Networks

Standard TNA assumes first-order Markov dynamics: the next state depends only on the current state. Higher-order analysis tests whether longer sequential memory exists — cases where what happened two or three steps ago changes the transition probabilities.

17.1 Model order selection

build_mogen() fits Markov models at increasing orders and selects the best via AIC/BIC:

data("human_cat", package = "Nestimate")
hon_net <- build_network(human_cat, method = "tna",
  action = "category", time = "timestamp", session = "session_id")
mg <- build_mogen(hon_net, max_order = 4)
summary(mg)
Multi-Order Generative Model (MOGen) Summary

  States: Command, Correct, Frustrate, Inquire, Interrupt, Refine, Request, Specify, Verify
  Paths: 1202 | Observations: 10563

 order layer_dof cum_dof    loglik      aic      bic selected
     0         8       8 -21525.46 43066.91 43125.03
     1        72      80 -20328.25 40816.51 41397.71
     2       605     685 -19334.29 40038.58 45015.18
     3      2061    2746 -16859.93 39211.85 59161.85
     4      2039    4785 -12426.49 34422.97 69186.54      <--

  Optimal order: 4 (by aic)
plot(mg)
Figure 29: Model selection: information criteria across Markov orders

17.2 Higher-Order Network (HON)

HON detects where in the state space higher-order dependencies exist — which transitions change depending on prior context:

hon <- build_hon(hon_net)
hon
Higher-Order Network (HON)
  Nodes:        903 (9 first-order states)
  Edges:        3324
  Max order:    5 (requested 5)
  Min freq:     1
  Trajectories: 1202

17.3 Path anomaly detection (HYPA)

HYPA identifies paths that occur significantly more or less often than expected under a hypergeometric null model:

hypa <- build_hypa(hon_net)
table(hypa$scores$anomaly)

normal   over  under
  2552    110     71

Over-represented paths — learned behavioral routines:

over <- subset(hypa$scores, anomaly == "over")

head(over)
                                            path
56      Command -> Command -> Correct -> Command
97      Request -> Command -> Correct -> Specify
98      Specify -> Command -> Correct -> Specify
158     Command -> Command -> Inquire -> Inquire
162     Specify -> Command -> Inquire -> Inquire
186 Interrupt -> Command -> Interrupt -> Command
                                 from      to observed expected    ratio
56      Command -> Command -> Correct Command       11 5.915291 1.859587
97      Request -> Command -> Correct Specify        5 1.394763 3.584838
98      Specify -> Command -> Correct Specify        8 4.358635 1.835437
158     Command -> Command -> Inquire Inquire        6 2.129505 2.817557
162     Specify -> Command -> Inquire Inquire        5 2.366116 2.113167
186 Interrupt -> Command -> Interrupt Command        7 1.481936 4.723551
    hypa_score anomaly
56   0.9821252    over
97   0.9969815    over
98   0.9664514    over
158  0.9938890    over
162  0.9668313    over
186  0.9998559    over

17.4 Simplicial visualization

plot_simplicial() renders higher-order pathways as smooth blobs over the network. Source states appear in blue, target states in red.

plot_simplicial(hon_net, max_pathways = 6,
  title = "Higher-Order Pathways")
Figure 30: Top 6 higher-order pathways (HON)

Dismantled view — one panel per pathway:

plot_simplicial(hon_net, max_pathways = 9,
  dismantled = TRUE, ncol = 3)
Figure 31: Individual pathway panels

HYPA anomalous pathways:

plot_simplicial(hon_net, method = "hypa", max_pathways = 9,
  dismantled = TRUE, ncol = 3)
Figure 32: Anomalous pathways detected by HYPA

18 Robustness

robustness() simulates network degradation under targeted and random node/edge removal:

plot_robustness(x = mat,
  measures = c("betweenness", "degree", "random"),
  n_iter = 50, seed = 42,
  title = "Node Removal Robustness")
Figure 33: Network robustness under targeted and random node removal

19 Saving and Exporting

19.1 Direct File Output 

# Save as PNG
splot(model, filename = "network.png", width = 8, height = 8, res = 300)

# Save as PDF
splot(model, filename = "network.pdf", width = 8, height = 8)

# Save as SVG
splot(model, filename = "network.svg", width = 8, height = 8)

19.2 Pipe-Friendly Saving 

model |>
  cograph(layout = "circle") |>
  sn_theme("dark") |>
  sn_nodes(size = 10) |>
  sn_save("network_dark.png", width = 8, height = 8, dpi = 300)

19.3 ggplot2 Export 

p <- model |>
  cograph(layout = "circle") |>
  sn_ggplot()

ggplot2::ggsave("network.pdf", p, width = 8, height = 8)

20 Complete Parameter Reference

splot() — All Parameters

Input & Layout: x, layout, directed, seed, theme, title, background, groups

Node Appearance: node_size, node_size2, node_shape, node_fill, node_border_color, node_border_width, node_alpha, scale_nodes_by, node_size_range

Labels: labels, label_size, label_color, label_position, label_fontface, label_fontfamily

Donut: donut_fill, donut_color, donut_inner_ratio, donut_shape, donut_show_value, donut_value_suffix, donut_value_digits, donut2_values, donut2_colors

Pie: pie_values, pie_colors

Edge Appearance: edge_color, edge_width, edge_size, edge_width_range, edge_scale_mode, edge_cutoff, edge_alpha, edge_style, edge_start_style, edge_positive_color, edge_negative_color, curvature, curves

Arrows: arrow_size, arrow_angle, show_arrows, bidirectional

Edge Labels: edge_labels, edge_label_size, edge_label_color, edge_label_bg, edge_label_position, edge_label_offset, edge_label_fontface, edge_label_style, edge_label_template

CI Underlay: edge_ci, edge_ci_scale, edge_ci_alpha, edge_ci_color, edge_ci_style

Weight / Threshold: threshold, minimum, maximum, weight_digits

Layout Control: rescale, layout_scale, layout_margin, aspect, loop_rotation

File Output: filename, filetype, width, height, res

TNA: i (group index for group_tna)

All Available Layouts

Built-in: "oval" (default), "circle", "spring", "grid", "random", "star", "bipartite", "groups"

igraph (two-letter codes): "kk" (Kamada-Kawai), "fr" (Fruchterman-Reingold), "drl", "mds", "lgl", "ni" (nicely), "tr" (tree), "st" (Sugiyama)

You can also pass a custom layout matrix (2-column matrix with x, y coordinates) or a layout function.

References

  • cograph: Modern R Package for Network Visualization. https://github.com/sonsoleslp/cograph

  • Saqr, M., López-Pernas, S., Törmänen, T., Kaliisa, R., Misiejuk, K., & Tikka, S. (2025). Transition Network Analysis: A Novel Framework for Modeling, Visualizing, and Identifying the Temporal Patterns of Learners and Learning Processes. In Proceedings of the 15th International Learning Analytics and Knowledge Conference (LAK ’25) (pp. 351–361). ACM. https://doi.org/10.1145/3706468.3706513

  • Tikka, S., López-Pernas, S., & Saqr, M. (2025). tna: An R Package for Transition Network Analysis. Applied Psychological Measurement. https://doi.org/10.1177/01466216251348840