Nexus

NEXUS is a plain-text data format for phylogenetic, taxonomic, and comparative biological data, introduced in 1997 by David R. Maddison, David L. Swofford, and Wayne P. Maddison in a paper in Systematic Biology. It grew out of a shared file convention developed from 1987 onward for the programs MacClade and PAUP, and became the de facto interchange standard across systematics software including PAUP*, MrBayes, Mesquite, SplitsTree, and Genetic Data Analysis. Files carry the extensions .nex or .nxs and begin with the magic token #NEXUS.

The defining design principle is modularity. A NEXUS file is a sequence of named blocks, each delimited by BEGIN blockname; and END;, so that a program can read the blocks it understands and silently skip the rest. The specification defines a set of public blocks: TAXA (the operational taxonomic units and their canonical order), CHARACTERS and the convenience block DATA (character-by-taxon matrices for morphological or molecular data), UNALIGNED, DISTANCES (distance matrices), CODONS, SETS (named CHARSET/TAXSET groupings), ASSUMPTIONS (weights, step costs, exclusion status), and NOTES for metadata. Within CHARACTERS, the MATRIX command holds rows of taxa against columns of character states, with optional TRANSPOSE and INTERLEAVE arrangements. Taxa, characters, and trees each carry an implicit ordering and may be referenced by name or by number.

Trees are stored in the TREES block as nested-parenthesis descriptions in the closely related Newick notation, for example Tree tree1 = ((Alpha,Beta),Gamma,Delta);, optionally with a TRANSLATE table mapping integers to taxon names. Syntactically NEXUS is token-based and case-insensitive, uses square brackets for comments, and terminates commands with semicolons.

In the graph and format-conversion landscape NEXUS is fundamentally a tree/matrix container rather than a general graph language, but its trees map directly onto the same rooted-tree model as Newick, phyloXML, and NeXML, and tools such as Biopython and ete3 read and write it. Its main weaknesses are the flip side of its flexibility: there is no formal machine-checkable grammar, non-compliant extensions have accumulated across programs so that files written by one tool may not parse in another, and it cannot cleanly express reticulate structures such as networks or horizontal gene transfer, nor certain character ambiguities. The XML-based NeXML was designed as a validatable successor to address exactly these interoperability gaps while preserving the NEXUS data model.