plexgraph

Interactive graph visualization for multilayer, temporal and hypergraph networks, from Python. Build a graph, call show(), and explore it in a WebGL viewer, inline in Jupyter or in a browser tab. It stays responsive on graphs with 100,000 nodes and hundreds of thousands of edges.

pip install plexgraph            # everything needed, including the viewer; no Node.js required
pip install "plexgraph[jupyter]" # plus the notebook widget (anywidget)
pip install "plexgraph[all]"     # plus pandas, networkx and the notebook widget

import plexgraph as pg

g = pg.Graph()
for i in range(60):
    g.add_node(i, team=f"t{i % 3}")
for i in range(60):
    g.add_edge(i, (i + 1) % 60)
    g.add_edge(i, (i * 7 + 3) % 60)

view = pg.show(g, node_color=pg.by_attribute("team", palette="tab10"),
               node_size=pg.size_by_degree((8, 18)), edge_curvature=0.15, return_handle=True)

view.style(node_shape="s")               # change the open viewer from Python
view.color_nodes([0, 1, 2], "crimson")   # paint specific nodes

Try it without writing code: python -m plexgraph demo.

What it does

• Three kinds of network as first-class data: layers (multiplex), time (contact sequences or intervals) and hyperedges, in one graph model.
• Loaders for edge lists, pandas DataFrames, networkx, GEXF and GraphML, and for temporal u v t files (pg.read_temporal_edgelist) with ISO dates, custom date formats or Unix timestamps.
• A layout that scales: streaming force layout with bounded memory, an overview that turns into individual nodes as you zoom, and communities that collapse and expand.
• networkx-style styling, live: colors, sizes, shapes, outlines, opacity, curved edges and labels, from attributes, degree, weights, time buckets or your own lists and dicts, changed while the viewer is open. The same controls are in the viewer's Appearance panel.
• Exploration tools: search, filters, click-to-select, shortest path, time slider and time ribbon, and export to PNG, JPG, SVG, HTML or PDF.

Documentation

• User guide
• Tour notebook with a runnable example of every feature
• Changelog

Notes

• The package is pure Python and the viewer is bundled, so there is nothing to compile or build.
• import plexgraph is the supported entry point. It is implemented by two packages installed alongside it, plexgraph_core (graph model and loaders) and plexgraph_bridge (viewer and styling). Do not install this next to the separate plexgraph-core or plexgraph-bridge development packages, which provide the same modules.
• Status: alpha. The API may change before 1.0.

MIT licensed.