Color by binding site or by x3dna-dssr motif. In motif mode each panel is colored by its own structure's motifs, so an apo–holo motif change shows up. Tip: click any nucleotide to fly the 3D viewer to that residue.
Each holo nucleotide is a cell in the feature track: apo→holo C3′ displacement (blue → red), a ● for binding-site residues at the live cutoff, and a x3dna-dssr motif band. Click a cell to fly the 3D view there.
Holo base pairs arch above the nucleotide axis (holo numbering), apo base pairs below: formed on binding (holo-only), disrupted (apo-only), unchanged (both) - non-canonical (Leontis–Westhof) pairs dashed. The formed/disrupted headline counts the canonical (Watson–Crick / wobble) pairs; any labile non-canonical ones are added separately. A change at a position where the apo and holo sequences differ (the ≥90%-identity cutoff, not 100%) is faded and marked ≠ - it may reflect the sequence difference rather than binding. Structure-derived (x3dna-dssr) + geometric (apo–holo alignment).
| Hydrogen bond | x3dna-dssr (PLIP where DSSR cannot read the ligand) |
| π-stacking · Halogen bond | PLIP |
| Metal coordination | PLIP: an ion bound to the ligand (ligand–ion, not RNA–ligand) |
| RNA–ligand proximity | each nucleotide within the cutoff, at its closest heavy-atom distance to the ligand; a distance, not an interaction call |
Highlight pocket marks the binding-site residues.
| Base pairing | x3dna-dssr base pairs of the binding-site chains, with the Leontis–Westhof and Saenger classes |
| Interaction pairs | each nucleotide within the cutoff: its distance to the ligand and the interaction type |
| Hydrogen bonds | RNA–ligand hydrogen bonds, atom by atom (x3dna-dssr; residue level from PLIP where DSSR cannot read the ligand) |
| Base-pair rewiring | the formed, disrupted and unchanged pairs of the Base-pair rewiring panel |
| Structural motifs | x3dna-dssr motifs of the binding-site chains; “In pocket” marks those touching the binding site |
CSV downloads the tab on screen.
| Holo & apo mmCIF | the two deposited structures ({holo}.cif, {apo}.cif) |
| Superposed apo | the apo superposed onto the holo ({id}_apo_sup.pdb) by
one rigid transform solved over every chain of the binding site at once — each pocket
chain contributes its own sequence-paired C3′ atoms to that single fit, so no chain is the
reference and none is merely carried along - via the sequence-anchored Kabsch fit (a one-chain fit
also tries US-align and keeps the lower RMSD) — the overlay
you see in 3D, trimmed to this pocket's chains exactly as the viewer receives it (the full deposited
apo is the {apo}.cif above) |
| Interaction contacts | the ligand–RNA contact atom pairs as CSV (contacts.csv) |
| Ligand | the ligand 3D coordinates (ligand.sdf) |
| Pair record | the full pair record — all metrics + metadata (pair.json) |
The bundle follows the apo structure on screen: the one this page was opened on
(named by the ?apo= in its URL when there is one) decides which superposed apo and pair
record are zipped.
The morph trajectory and 2D structures aren't zipped — grab them from the individual
file links on the card (or the API).
in the holo only
in the apo only
in both (grey)
Each chain becomes one axis (the lower one drawn 3′→5′) and every cross-strand pair a rung, coloured formed / disrupted / unchanged like the single-chain arcs.
A Needleman–Wunsch alignment of the full declared (SEQRES) sequences: identical columns ÷ all columns, so every mismatch and gap counts, ends included. Unmodeled residues do not lower it. Use it for a full-length match.
The upstream BLAST pident, measured only inside the aligned region, so terminal overhangs are ignored. Every served pair is ≥90% over ≥90% coverage.
A multi-chain pocket shows one pair of numbers per chain, on that chain’s header: each box reports its own chain, never an aggregate over the pocket. The value published for the pair – table, filters, downloads – is the ligand’s own (auth) chain’s, not the lowest of them. (It was the minimum over the chains until 2026-07-28.)
Resolved in 3D. Hover for its PDB number; can be underlined as a binding-site residue.
In the deposited sequence but disordered, so it has no coordinates: no number, no click target, never underlined. A faded position is not a sequence difference.
The position does not exist in that construct.
A real substitution; it counts against the Global identity.
Both sides are the same canonical base, so the column is a match and costs the
Global identity nothing – but the residues actually deposited there differ,
e.g. holo U against apo PSU (pseudouridine). Hover the column
for both deposited names. Not red, because it is a chemical modification and not a
substitution.
N = parent base undeterminedA modified residue normally shows its parent base (pseudouridine reads U). When it
has no A/C/G/U parent it reads N – hover the N to see which
residue it is, e.g. “Modified residue: 6FU · Canonical parent base:
undetermined”; a residue with no base at all reads “abasic / unresolved”. Two
Ns match only if they are the same residue (LCC vs LCC); ZTH vs 8RJ is a red
column.
The apo structure on screen, whichever card opened the page (with or without
?apo=), is indexed on its own declared sequence, so every entry reads on the same
basis. (An upload on the Query page has no deposited declared
sequence, so its box aligns modeled residues only.)
A multi-chain pocket has one box per chain, each scored on its own chain. The value published for the pair is the ligand’s own (auth) chain’s – not a minimum over the chains.
a column present in one row and not the other. Here it can mean either a true insertion/deletion or simply a residue that was not resolved (disordered) in that crystal – the two are indistinguishable without the declared sequence.
the two modeled residues differ. Because disorder can shift the alignment register, a red column here is not guaranteed to be a real substitution.
a c g u ta modified nucleotide as x3dna-dssr letters it. It is compared case-insensitively, so a modified/unmodified pair is a match, not a red column.
Every pair served from the database itself uses the declared-sequence box instead – see its ? for the three-class reading (modeled / declared-but-unmodeled / true indel).