How to visualize communities from a list in igraph python

Question

I have a community list as the following list_community. How do I edit the code below to make the community visible?

from igraph import *

list_community = [['A', 'B', 'C', 'D'],['E','F','G'],['G', 'H','I','J']]
list_nodes = ['A', 'B', 'C', 'D','E','F','G','H','I','J']
tuple_edges = [('A','B'),('A','C'),('A','D'),('B','C'),('B','D'), ('C','D'),('C','E'),
              ('E','F'),('E','G'),('F','G'),('G','H'),
              ('G','I'), ('G','J'),('H','I'),('H','J'),('I','J'),]

# Make a graph
g_test = Graph()
g_test.add_vertices(list_nodes)
g_test.add_edges(tuple_edges)

# Plot
layout = g_test.layout("kk")
g.vs["name"] = list_nodes
visual_style = {}
visual_style["vertex_label"] = g.vs["name"]
visual_style["layout"] = layout
ig.plot(g_test, **visual_style)

I would like a plot that visualizes the community as shown below.

I can also do this by using a module other than igraph. Thank you.

Not strictly a duplicate (so not flagging as such), but perhaps one of these questions would be helpful since it sounds like you're open to using networkx and matplotlib: [(1)](https://stackoverflow.com/questions/40941264/how-to-draw-a-small-graph-with-community-structure-in-networkx) and [(2)](https://stackoverflow.com/questions/43541376/how-to-draw-communities-with-networkx) — Frodnar, Nov 28 '22 at 09:22
@Frodnar, Thanks for your suggestion. These tips are helpful, but I don't think they address the case where communities overlap, as they are colour-coded by a node. In my example, node G spans two communities. — mimi, Nov 29 '22 at 05:18
@mimi In order to represent overlapping clusters, you can use the `VertexCover` object of igraph, and you can plot it using `plot(cover, mark_groups=True)`. It appears that one needs to use numeric vertex _indices_ instead of vertex _names_ to specify the groups. Since I am not sure what is the simplest way to deal with this, I will leave giving an answer to others. — Szabolcs, Nov 29 '22 at 11:55

score 3 · Answer 1 · answered Nov 30 '22 at 13:34

In igraph you can use the VertexCover to draw polygons around clusters (as also suggested by Szabolcs in his comment). You have to supply the option mark_groups when plotting the cover, possibly with some additional palette if you want. See some more detail in the documentation here.

In order to construct the VertexCover, you first have to make sure you get integer indices for each node in the graph you created. You can do that using g_test.vs.find.

clusters = [[g_test.vs.find(name=v).index for v in cl] for cl in list_community]
cover = ig.VertexCover(g_test, clusters)

After that, you can simply draw the cover like

ig.plot(cover,
        mark_groups=True,
        palette=ig.RainbowPalette(3))

resulting in the following picture

score 1 · Accepted Answer · answered Nov 29 '22 at 10:11

Here is a script that somewhat achieves what you're looking for. I had to handle the cases of single-, and two-nodes communities separately, but for greater than two nodes this draws a polygon within the nodes.

I had some trouble with matplotlib not accounting for overlapping edges and faces of polygons which meant the choice was between (1) not having the polygon surround the nodes or (2) having an extra outline just inside the edge of the polygon due to matplotlib overlapping the widened edge with the fill of the polygon. I left a comment on how to change the code from option (2) to option (1).

I also blatantly borrowed a convenience function from this post to handle correctly sorting the nodes in the polygon for appropriate filling by matplotlib's plt.fill().

Option 1:

Option 2:

Full code:

import networkx as nx
import matplotlib.pyplot as plt
import numpy as np
from matplotlib import cm

def sort_xy(x, y):

    x0 = np.mean(x)
    y0 = np.mean(y)

    r = np.sqrt((x-x0)**2 + (y-y0)**2)

    angles = np.where((y-y0) > 0, np.arccos((x-x0)/r), 2*np.pi-np.arccos((x-x0)/r))

    mask = np.argsort(angles)

    x_sorted = x[mask]
    y_sorted = y[mask]

    return x_sorted, y_sorted

G = nx.karate_club_graph()

pos = nx.spring_layout(G, seed=42)
fig, ax = plt.subplots(figsize=(8, 10))
nx.draw(G, pos=pos, with_labels=True)

communities = nx.community.louvain_communities(G)

alpha = 0.5
edge_padding = 10
colors = cm.get_cmap('viridis', len(communities))

for i, comm in enumerate(communities):

    if len(comm) == 1:
        cir = plt.Circle((pos[comm.pop()]), edge_padding / 100, alpha=alpha, color=colors(i))
        ax.add_patch(cir)

    elif len(comm) == 2:
        comm_pos = {k: pos[k] for k in comm}
        coords = [a for a in zip(*comm_pos.values())]
        x, y = coords[0], coords[1]
        plt.plot(x, y, linewidth=edge_padding, linestyle="-",  alpha=alpha, color=colors(i))

    else:
        comm_pos = {k: pos[k] for k in comm}
        coords = [a for a in zip(*comm_pos.values())]
        x, y = sort_xy(np.array(coords[0]), np.array(coords[1]))
        plt.fill(x, y, alpha=alpha, facecolor=colors(i), 
                 edgecolor=colors(i), # set to None to remove edge padding
                 linewidth=edge_padding)

How to visualize communities from a list in igraph python

2 Answers2