// Single-seq arm · high RMSD · strict@1 anyway

Backbone RMSD 14 Å. Rank-1 cluster still strictly recovers the binding site.

2HOJ is the inverse of 2GIS. The predicted backbone is the worst-fitting in the v0.2 benchmark — RMSD 13.95 Å against the deposited co-crystal — yet the rank-1 cluster lands at 53% binding-site overlap: STRICT@1. The TPP-binding pocket geometry is preserved within the prediction even as the global fold drifts away from native.

Read 2HOJ alongside 2GIS (near-perfect fold, only NEAR@1) and 5C45 (poor fold, NEAR@1 anyway). Together the three cases pin down what backbone RMSD does and does not tell you about whether the pipeline recovered a binding site: it is a useful diagnostic, not the quality metric. The metric that maps to customer usefulness is binding-site overlap.

RMSD
14 Å
global
overlap
53%
rank-1
verdict
strict@1
local geometry

Numbers from locked v0.2 benchmark ·  strict@K / near@K definitions

Example output — non-customer demo

RNA pocket discovery

thi-box TPP riboswitch · 2HOJ

Bound by thiamine pyrophosphate (PDB ligand TPP). 83 nt RNA target.

v0.2 scope: cleft-binding RNA ≤500 nt|Pre-pilot screen: PASS
Download:↓ Ensemble (PDB)↓ Pocket data (JSON)↓ Full report (PDF)
// pre-pilot screenPASSsingle-seq armmin_id 0.761 · 0.0% homologs at 70–80% identityNo diverse-tail homologs — single-seq prediction is the appropriate arm
Sequence length
83 nt
Conformers sampled
5
Structure pLDDT (mean)
0.754
Pocket clusters
5 / 13
passing persistence floor
thi-box TPP riboswitch predicted structure with top-3 candidate pockets highlighted
Predicted structure, top-3 pockets highlighted.Rank 1Rank 2Rank 3

Top-3 candidate pockets

ranked by persistence × binding-residue stability
#1Cluster 1persistence 100%
Geom. score
7.00
persistence × intersected
Residues (∩)
7
every frame
Residues (∪)
10
union, all frames
residues (union): 11, 12, 31, 32, 34, 35, 37, 50, 64, 65
residues (∩ all frames): 11, 12, 31, 32, 34, 35, 65
benchmark9/17 binding-site residues (53%) — strict recovery
#2Cluster 9persistence 60%
Geom. score
3.60
persistence × intersected
Residues (∩)
6
every frame
Residues (∪)
10
union, all frames
residues (union): 50, 51, 52, 53, 64, 66, 67, 68, 69, 70
residues (∩ all frames): 50, 51, 52, 64, 67, 68
benchmark8/17 binding-site residues (47%) — near recovery
#3Cluster 4persistence 40%
Geom. score
3.20
persistence × intersected
Residues (∩)
8
every frame
Residues (∪)
9
union, all frames
residues (union): 46, 47, 49, 50, 70, 71, 72, 73, 74
residues (∩ all frames): 46, 47, 49, 50, 70, 71, 72, 73
benchmark3/17 binding-site residues (18%)
Rotate the structure yourself

Cartoon backbone of the predicted reference frame. Top-3 pocket residues highlighted as licorice; centroid spheres mark each cluster's geometric centre across the ensemble. Hover a card to isolate that pocket. Toggle the experimental TPP overlay to see where the co-crystal ligand sits relative to the rank-1 cluster.

Sequence with pocket residues highlighted

GCGACUCGGGGUGCCCUUCUGCGUGAAGGCUGAGAAAUACCCGUAUCACCUGAUCUGGAUAAUGCCAGCGUAGGGAAGUCGCA
Rank 1 pocket residuesRank 2 pocket residuesRank 3 pocket residuesknown binding-site residue

Methods summary

v0.2 detects cavities on the predicted 3D structure using RNA-tuned fpocket parameters (consistent with the published fpocketR approach, Veenbaas et al. PNAS 2025), samples a 5-frame ANM conformational ensemble, and ranks pockets by structural persistence weighted by binding-residue stability (score = persistence × n_residues_intersected). The cross-frame geometric ranker is the load-bearing contribution: on a 7-target cleft-binder benchmark, RNA-tuned detection alone recovers 0 of 7 at strict@1; the ensemble + ranker lifts recovery to 3 of 7 strict@1 and 6 of 7 near@1. Druggability assessment itself is left to your medicinal-chemistry workflow; we provide the geometric metadata and conformational stability metrics as inputs to it. Computational predictions only — experimental validation is required before use in drug development.

// rank-1 binding-site overlap, this target

fpocketR single-frame
0%NEITHER
→
v0.2 ensemble + ranker
53%STRICT

+53 pp · NEITHER → STRICT

Read full methodology →

Binding-mode caveat

Pipeline detects cleft-shaped binding pockets. Groove-binding modes and shallow surface-deformation binding may be missed. Contact us if your target's binding mode is groove-mediated.

Computational predictions only. Experimental validation required before drug-development use.