SBGN-ML

SBGN-ML is the XML-based interchange format for the Systems Biology Graphical Notation (SBGN), a standardized visual language for representing biological networks such as metabolic pathways, signalling cascades, and gene-regulatory interactions. SBGN itself was initiated at a 2006 Tokyo meeting led by Hiroaki Kitano and drawing on Kurt Kohn's Molecular Interaction Maps, and is developed by a community of biochemists, modelers, and computer scientists coordinated under COMBINE (Computational Modeling in Biology Network). SBGN defines three complementary sub-languages: Process Description (PD), which shows temporal sequences of molecular events; Entity Relationship (ER), which captures the relationships an entity participates in independent of time; and Activity Flow (AF), which depicts the flow of influence between activities. SBGN-ML version 0.3 (namespace http://sbgn.org/libsbgn/0.3) is the first release whose schema covers all three languages in a single format.

The format's design goal is to capture both the geometry and the biological meaning of a map so that rendering software need perform no complex layout computation. A document is rooted in an sbgn element containing one or more map elements, each tagged with a language attribute. A map holds glyph and arc elements: every glyph carries a class attribute encoding its SBGN meaning, an id for cross-reference, and a mandatory bbox (bounding box) giving its position and size. Glyphs may nest, carry a label, state, or clone, and expose port elements as named anchor points. An arc has a class, a source and target (a glyph or port), and an explicit geometric path built from straight segments and quadratic or cubic Bezier curves. An optional render extension, adapted from SBML's, adds color and style definitions.

Reference tooling is provided by libSBGN, distributed as C++ and Java libraries, with community bindings such as pysbgn and libsbgn-python; it reads, writes, and validates SBGN-ML and can convert to and from related standards. SBGN-ML is complementary to, rather than a replacement for, the model-oriented SBML, and relates to pathway representations including BioPAX, CellDesigner, GPML, and KGML. Editors and viewers include SBGN-ED, Newt, and Cytoscape via cy3sbml.

Its strengths are precise, unambiguous semantics and self-contained layout that makes maps portable and faithfully renderable. Its limitations are practical: software support historically concentrated on Process Description, adoption remains narrower than SBML's, and authors must learn a substantial standardized symbol vocabulary.