new PDB depositions vs. their blind AlphaFold predictions — a running test of "is folding solved?"

13GJ_A

IST1 homolog · P53990 · RCSB 13GJ · AF-P53990-F1 (v6)

0.968
TM-score
0.9747
lDDT
0.932
Cα-RMSD Å
94.33
mean pLDDT
0.052
FRAUD score
100.0%
novelty (max id)

Per-domain breakdown

SourceDomainRangeResiduesTMRMSD Åmean Cα ΔName
PAEPAE:1-3641-364 1830.968 0.9320.834

All metrics

Global fold agreement

TM-score (norm. experiment)0.968
TM-score (norm. model)0.494
TM-score (norm. shorter)0.968
TM-score (norm. longer)0.494
Cα-RMSD (Å)0.932
backbone-RMSD (Å)0.935
all-atom-RMSD (Å)1.428
core-RMSD (Å)0.91
core fraction0.989
GDT_TS93.03
GDT_HA73.63
MaxSub0.968
structural overlap (3.5 Å)1.0

Local, superposition-free

lDDT0.9747
contact-map Jaccard0.8299
contact precision0.8703
contact recall0.9471
distance-matrix mean Δ (Å)0.311
CAD-score (approx)0.2753

Backbone & secondary structure

SS agreement Q3 (%)89.62
mean Δφ (°)7.1
mean Δψ (°)7.7
torsion within 30° (frac)0.9727
Rg experiment (Å)20.04
Rg model (Å)19.96
ΔRg (Å)0.09

Confidence calibration

mean pLDDT94.33
pLDDT↔lDDT Pearson0.5266
pLDDT↔lDDT Spearman0.2818
PAE↔observed Pearson0.2641
PAE overconfident frac0.0
mean PAE (Å)4.668
mean observed error (Å)0.309

Context & headline

coverage of model0.5027
coverage of experiment1.0
seq identity aligned (%)100.0
confidently-wrong residue frac0.0
FRAUD score0.052

Deposited 2026-05-05 · released 2026-07-08 · EM · 2.0 Å · closest pre-cutoff chain: 3FRR_1