Which file formats Nanome opens, and how it handles bonds, crystals and trajectories

Nanome opens PDB, mmCIF, SDF, SMILES, XYZ, PQR and PDBQT structures, Maestro, MOE and PyMOL session files, GRO, XTC, TRR and DCD molecular dynamics trajectories, and DX electrostatic maps, according to its supported file formats page. MOL and MOL2 files load through a MARA tool. A .cif opens through the mmCIF/PDBx reader, a multi-block .xyz plays as an animation as of 2.6.0, and MD trajectories play frame by frame in the browser and in XR headsets. Export is narrower: one frame at a time, as PDB, SDF or SMILES.

That reader handles the Protein Data Bank's dialect of CIF, and the formats page describes no symmetry expansion or crystal packing, so a small-molecule crystal CIF, or a packed lattice built from one, goes through a crystallography tool such as Mercury and arrives in Nanome as a PDB.

What Nanome opens, next to Nanome Classic

The Nanome column follows the formats page, current through 2.6.0. The Classic column follows the Classic answer in Nanome's FAQ, which names PDB and SDF as Classic's most stable formats.

The formats page routes .bcif, .mcif, .sd, .mol and .mol2 through a MARA tool that converts or loads them, and compressed .gz and .maegz files load as of 2.6.1. On the way out, PDB and SDF come from every export surface and SMILES from the WorkspaceAPI and MCP export_entry route (export notes). There's no trajectory export and no portable workspace file.

Three things stay off the list: electron density maps (CCP4, MRC, DSN6), native LAMMPS trajectories, and 64-bit CHARMM or fixed-atom DCD files (not supported, DCD limits). Classic's FAQ lists .ccp4 and .dsn6 maps. In current Nanome, map work stays in Coot or ChimeraX, the split laid out in molecular visualization for structural biologists.

Getting molecules into a workspace

• By PDB ID. The web app loads structures from the RCSB by ID, and inside a headset the Download from RCSB tool does the same.
• By upload. Files drag and drop into a workspace in the web app. From inside XR, other file types go in through the web app and sync to the headset, as the Main Panel docs describe.
• By name or SMILES. Retrieving a compound's SMILES and turning SMILES into a 3D structure are both on the MARA features list.
• By hand. The Builder draws a molecule in 3D from atoms, rings and functional groups, and can minimize it with MMFF94, MMFF94s or UFF.

Classic's hardware guide strongly recommends the Vault plugin for loading files on standalone headsets. From Vault to workspaces describes Vault's cloud folders and how 2.0 replaced them with workspaces.

How large a file can be

On a headset, the practical limit is frame rate, and the session guide ties frame rate to how many atoms are on screen. It rates a Quest 3 comfortable at about 7,500 shown atoms and over budget at 15,000, counting what's displayed rather than what's loaded. Per-device figures are in which VR headsets work with Nanome.

Frame trajectories carry a 2000-frame cap, and an attached file's atom count has to match its host model's (how animation works). The 2.6.0 notes add frame limits that take atom count into account, and they describe a load manager that hides very heavy structures to protect frame rate.

Multi-model files have no published cap. Every model in one carries a full topology, while a frame trajectory keeps one topology and stores only coordinates per frame (how animation works). A run with hundreds of snapshots travels lighter as a topology plus an XTC or DCD, or split into shorter files.

Why bonds flicker during playback

A PDB file lists bonds for HET groups and for links the standard residue table doesn't cover, while bonds inside standard residues come from that table, outside the file, per the wwPDB format guide. Every viewer works out the rest itself, Nanome included.

In a multi-MODEL PDB, each playback step is a whole model with its own topology (how animation works), so bonds get worked out again for every model. Where a viewer guesses bonds from elements and positions, as the LAMMPS docs note many do, a bond stretched far enough in one snapshot can drop out and return in the next, which shows up as flicker.

SDF records carry their own bond block, so a multi-record SDF keeps the bonds the file states. For MD, the steadier route is a frame trajectory: the topology loads once, and each frame of an XTC, TRR or DCD moves coordinates on that one shared model.

Trajectories, vibrational modes and LAMMPS output

Model-index playback steps through whole models in a multi-MODEL PDB or mmCIF, a multi-record SDF or a multi-block XYZ. Frame-trajectory playback steps through coordinate sets on one model, from a standalone GRO or an attached XTC, TRR or DCD. Both run on the same play, pause and scrub controls (how animation works).

Multi-block XYZ playback was fixed in 2.6.0, so an XYZ that wouldn't play on 2.5 is worth reloading. That suits a vibrational mode written as one XYZ block per displacement step. A normal mode across a whole protein does better as a multi-MODEL PDB, since XYZ holds only elements and coordinates, with no residues for a cartoon to follow. ProDy's traverseMode generates conformations along a mode for this kind of animation.

Surfaces switch off during trajectory playback, since recomputing one every frame is still too slow (formats page). Stepping through frames one at a time gets a fresh surface per frame as of 2.6.0. Playback stays in sync for everyone in the session, as the v2.5 release post describes, and the MARA features page lists RMSD across a trajectory for the numbers. The comparison with VMD and other desktop MD tools is in software for visualizing molecular dynamics trajectories.

LAMMPS dump files don't load, and the formats page names XYZ as the bridge. LAMMPS writes XYZ with its xyz dump style, and the LAMMPS docs suggest dump_modify element so viewers get element names in place of numeric atom types. The formats page says nothing about coarse-grained models such as SIRAH, whose beads stand for groups of atoms, so a back-mapped atomistic trajectory is the safer input.

Crystal structures from a CIF

A small-molecule CIF stores the unit cell, the space group and the atoms of the asymmetric unit, usually as fractional coordinates, and the rest of the crystal comes from applying the symmetry. Nanome's formats page lists .cif only under mmCIF/PDBx, the dialect the Protein Data Bank uses for macromolecules, and describes no symmetry expansion or packing step. A small-molecule CIF is therefore safest converted to PDB before upload, and a packed lattice gets built in a crystallography tool first.

The Mercury user guide covers both:

1. Packing. Calculate, then Packing/Slicing, with the Pack box ticked, sets ranges along a, b and c. A range of 0 to 1 shows one unit cell, and wider ranges add more.
2. Saving. Save As in PDB format writes every molecule on display, and PDB opens in Nanome directly.

Before upload, giving each copy its own chain ID or residue number in the PDB lets Nanome's select by chain and chain coloring pick out one copy at a time.

Every copy adds atoms, and against the session guide's 7,500-atom comfort figure for a Quest 3, a small block of cells suits a standalone headset better than a wide slab.

Classic's CIF support was limited (FAQ). Converting to PDB first sidesteps the CIF reader entirely: Mercury handles small-molecule CIFs, and MARA's mmCIF-to-PDB tool handles mmCIF. The same tool library searches and downloads from the Materials Project for inorganic crystals. XYZ is the other way in: a supercell exported as XYZ from crystallography software loads directly, though it carries no chains or residues to select by. Teaching chemistry and biology in VR builds its crystal lesson on XYZ supercells.

Molecules for teaching

Classroom structures come from the same places as research ones. Nanome's academic page points to the RCSB, PubChem and DrugBank, plus structures students build themselves, and MARA turns a SMILES string into a 3D model. Once a structure is in, picking representations is covered in the PDB viewing guide.

FAQ

Does Nanome read .xyz files?
Yes. Current Nanome opens .xyz directly, and multi-block files play as animations since 2.6.0, per the supported formats page. Nanome Classic lists .xyz in its FAQ as well.

Why won't my CIF load?
The likeliest cause is a parse failure. Current Nanome reads .cif as mmCIF/PDBx, and Classic's CIF support was limited (FAQ), so converting to PDB, in Mercury for a small-molecule CIF or with MARA's mmCIF-to-PDB tool for mmCIF, is the dependable fix. In current Nanome, the 2.6.0 Loading Report flags import problems and fixes the ones it can.

Can Nanome open MOL2 files?
Yes, through a MARA tool. The formats page routes .mol2 that way, and MARA's tool library includes a Mol2 to SDF and PDB converter. Nanome Classic opened .mol2 directly.

Can Nanome pack a crystal the way Mercury does?
No packing or symmetry expansion appears on the formats page, so the lattice gets built in Mercury and saved as PDB, with each copy on its own chain or residue number.

Can Nanome open LAMMPS output?
The native dump files don't load. LAMMPS can write the same run with its xyz dump style, and the formats page names XYZ as the bridge.