How to Use RMSD Calculator

RMSD Calculator measures structural deviation between every member of a Reference Group and every member of a Mobile Group. If the reference group contains R structures and the mobile group contains M structures, the result contains R multiplied by M comparisons. This all-against-all design is useful for comparing predicted models with several experimental references, evaluating conformational ensembles, screening mutant structures against canonical states, or measuring every frame representative against every cluster center.

The service accepts single-model PDB, CIF, and mmCIF files containing proteins, nucleic acids, or mixed protein-nucleotide assemblies. Small molecules, other hetero records, and waters are removed during input preparation because this workflow reports polymer backbone RMSD. The calculation uses corresponding backbone atoms rather than all heavy atoms, reducing sensitivity to side-chain rotamers while retaining the large-scale structural differences most users intend to measure.

RMSD Calculator assumes known atom correspondence. Compared structures must have matching chain identifiers, residue identifiers, insertion codes, and backbone atoms. It does not perform sequence alignment, residue remapping, or optimal chain assignment. Use USalign Structural Alignment when comparing remote homologs, differently numbered structures, or models whose residue correspondence must first be discovered.

For each reference-mobile pair, the service matches protein and nucleotide backbone atoms by chain ID, residue ID, insertion code, and atom name. With Align Structures enabled, it first applies a Kabsch least-squares superposition to the mobile coordinates and then calculates the root-mean-square distance between matched backbone atoms. This removes differences caused only by global rotation or translation and is the appropriate default for most structural similarity questions.

When Align Structures is disabled, RMSD is calculated directly from the submitted coordinates. This mode is meaningful when structures already occupy the same coordinate frame, such as snapshots from one restrained simulation, models transformed by a shared external workflow, or poses intentionally positioned relative to a fixed reference. A large unaligned RMSD can reflect simple translation or rotation rather than an internal conformational difference, so disabling superposition should be a deliberate choice.

Each reference is compared independently with every mobile structure. Mobile coordinates are restored before each pair so a superposition against one reference cannot influence the next comparison. The service does not export transformed structures; use Align Structures when aligned coordinate files are also needed.

Results are written to rmsd_values.csv with Reference Structure, Mobile Structure, and Backbone RMSD (A) columns. Values are reported in angstroms and rounded to three decimal places. RMSD is most informative when interpreted with the selected atom correspondence and the size and flexibility of the structures: the same numerical value can represent a localized loop movement in a large protein or a substantial global change in a small domain.

• Complete cross-group comparison: Calculate every reference-mobile pair automatically instead of preparing individual two-structure jobs.
• Optional optimal superposition: Use Kabsch-aligned RMSD by default or preserve a shared coordinate frame when the experiment requires it.
• Protein and nucleotide support: Compare proteins, DNA, RNA, and mixed polymer assemblies through one consistent backbone-matching workflow.
• Independent pair handling: Restore mobile coordinates between comparisons so each RMSD is calculated from the original submitted structure.
• Analysis-ready output: Receive one tidy CSV table that can be sorted, filtered, plotted as a heatmap, or joined with model metadata.