XYZ

XYZ is a minimal, human-readable text format for describing molecular geometry as a list of atoms with Cartesian coordinates. Originating around 1990 as part of the XMol visualization suite at the Minnesota Supercomputer Center, it has become one of the most ubiquitous interchange formats in computational chemistry precisely because of its simplicity. A file consists of a first line giving the atom count, a second free-form comment/title line, and one line per atom holding an element symbol (occasionally an atomic number) followed by x, y and z coordinates, conventionally in angstroms and separated by whitespace. There is no formal standard, so parsers must tolerate small dialect differences.

The defining trait of XYZ is what it omits: it records only atomic positions, not bonds, bond orders, formal charges, or any connectivity. Downstream tools that need a molecular graph must perceive it themselves, typically by comparing interatomic distances against sums of covalent radii. This makes XYZ trivial to generate and read but lossy relative to connection-table formats such as Molfile and SDF, or the richer XML-based CML. Multiple frames can be concatenated in one file — each with its own atom-count and comment line — to store optimization paths or molecular-dynamics trajectories, and the count may even differ between frames.

Extended XYZ (extxyz), driven by the Atomic Simulation Environment (ASE) and the libAtoms project, layers structured metadata onto the otherwise wasted comment line using key=value pairs. It encodes the periodic unit cell (Lattice), boundary conditions (pbc), and a Properties string that names and types the per-atom columns, so additional data like forces, velocities, or magnetic moments can travel alongside positions. Extxyz is the de facto standard for atomistic simulation and machine-learning interatomic potential datasets.

In the graph-and-format-conversion landscape XYZ sits at the geometric end of chemistry formats: it carries the spatial embedding of an atomic graph but leaves the edges implicit. It is widely readable and writable by tools such as Open Babel, RDKit, NetworkX (via chemistry bridges), and essentially every molecular visualizer (Jmol, VMD, Avogadro). Its strengths are portability, editability, and low overhead; its limitations are the absence of connectivity and any authoritative specification, which can cause silent inconsistencies between implementations.