KNeXT

KNeXT (KEGG NetworkX Topological parser) is a Python 3 command-line tool and library that downloads, parses, and converts pathway data from the Kyoto Encyclopedia of Genes and Genomes (KEGG) into tabular network files suitable for graph analysis. It was written by Everest Uriel Castaneda and Erich J. Baker (Baylor University and Belmont University) and released under an MIT license, with a 2024 description paper in Frontiers in Genetics. It is built directly on NetworkX (NetworkX) rather than layered on top of other KEGG parsers such as KEGGconverter or KEGGgraph (KEGGgraph).

The tool consumes KEGG Markup Language files (KGML), the XML dialect KEGG uses to encode annotated pathway maps. KNeXT centers on a problem it calls the "topological environment": KGML reuses the same gene/accession identifiers across disconnected neighborhoods of a map, so naive parsers collapse those separate regions into one graph and fabricate gene-gene edges that have no biological basis. KNeXT instead appends terminal modifiers to duplicated entry identifiers, preserving each occurrence's independent context, and uses shortest-path traversal plus "compound propagation" (dropping chemical intermediates while keeping the gene-to-gene connectivity they mediated) to reconstruct edges faithfully.

Output is a source-target edge list in tab-separated values (TSV), with columns for the two entries, the KGML relation/interaction type (activation, inhibition, phosphorylation and similar), and weights. A --graphics flag additionally emits a coordinate dictionary of node x-y positions so the original KEGG map layout can be reproduced in NetworkX visualizations. The CLI exposes a handful of verbs: get-kgml to download maps for a species, genes to build gene-only networks, mixed to retain genes, compounds, and pathway nodes together, and convert to remap KEGG identifiers to UniProt or NCBI accessions. It depends on NetworkX, pandas, requests, and typer/click.

KNeXT is deliberately narrow: it reads only KGML and does not ingest SBML (SBML), BioPAX (BioPAX), or SBGN, and it does not itself perform visualization or downstream statistics. Its TSV output is a generic edge list, so pairing it with converters or importers is needed to reach richer interchange formats like GraphML or SIF for Cytoscape (Cytoscape). The authors note that terminal modifiers must be stripped before external ID mapping, and that integration with downstream tooling remains an area for further work. For KEGG-to-network extraction where topological fidelity matters, however, it fills a specific and well-argued gap.