Graph software

Python

• NetworkX : the standard, pure Python
• igraph : C++ core with Python and R bindings, looks nice but GPL license
• graph-tool : new-comer, based on Boost Graph Library , looks nice but GPL license

General commentary on Python graph APIs .

R

• graph and Rgraphviz (Bioconductor)
  • Rare but usable Artistic 2.0 license
  • Arbitrary node/edge attributes (vignette )
  • Graphs are immutable (modifying nodes/edges creates new graph)
  • Graphs are simple but there is an “experimental” MultiGraph
• igraph (CRAN), GPL
  • Edge list representation, implemented in C, but conversion to/from adjacency list
  • Arbitrary graph/node/edge attributes
  • Only “limited support” for multigraphs
• network (CRAN), GPL
  • Edge list representation, implemented in C
  • Arbitrary graph/node/edge attributes
  • “Native support” for multigraphs and even hypergraphs

Julia

• LightGraphs.jl : the new standard, inspired by NetworkX
• Graphs.jl : the old standard but now phased out, inspired by the Boost Graph Library
• MetaGraphs.jl : extends LightGraphs with graph/vertex/edge properties
• Graft.jl : interesting but immature, emphasis on queries and data analysis

JavaScript

Data structures, layout, and visualization for graphs, partly taken from the StackOverflow list :

• Viz.js : GraphViz in JS via Emscripten
• JSNetworkX : Port of NetworkX to JS
  • Partial port: core functionality in place, most algorithms missing
  • Not maintained: Last update in 2015
• Cytoscape.js : active, mature, well-documented
  • Some algorithms, mostly visualization
  • No typings for TypeScript
  • R interface through RCyjs
• KLayJS : KIELER ’s layout algorithms for JS
  • Plugins for Cytoscape , NoFlo
  • Superseded by elkjs , ELK ’s layout algorithms for JS

Visualization for dataflow programming. These libraries tend to come equipped with useless execution engines but also to visualize port graphs as first-class entities.

• Rete.js : visual programming framework (GitHub , HN )
  • Flow graph editor inspired by Yahoo Pipes
  • UI framework agnostic; uses Vue by default with no official React support
• Lightgraph.js : graph node engine and editor (GitHub )
  • Renders in HTML5 Canvas
  • Not possible to embed? (#58 )
• NoFlo : flow-based programming for JS (GitHub )
  • Basis for Flowhub IDE
  • Visually attractive flow graph editor in the-graph component
  • Nodes are fixed size, which has pros and cons (#273 , #275 )
  • Not possible to embed? (#160 )

Commercial libraries:

• yFiles for HTML : Includes Graphviz-grade layouts
• Rappid : Build diagram editing tools for the web
  • Based on open-source library JointJS (GitHub )

XML formats:

• GraphML (primer ): widely supported, default in yEd
  • Advanced features: hypergraphs, nested graphs, ports
• GEXF (primer ): default in Gephi
  • Advanced features: nested graphs, dynamics (discrete and continuous time)
  • Gephi’s overview of graph data formats

JSON and related formats:

• GML (report ): custom syntax based on hierarchical key-value pairs
• GraphSON : only supported by TinkerPop3
• JGF (schema ): JSON Graph Format (GitHub )
  • Proposed standard, currently little or no uptake
• JSON KGraph : KIELER ’s graph format
  • Advanced features: nested graphs, ports, layout information
  • Supported by KLayJS (see above)
  • Related to, possibly identical with, ELK’s JSON format and supported by elkjs (again, see above)

The JSON formats are all about the same, with a list of nodes and a list of edges as top-level properties of a graph object.