wwPDB July 15, 2026 - Falsifier Verdicts
One table per prediction pathway. The only question answered here: did the terrain trigger the declared falsifier?
How to read this sheet. Each table below states the exact preregistered prediction and the exact preregistered falsifier - both written before the terrain existed. It then states what the terrain returned. The only adjudication question is binary: did the terrain produce the declared failure condition?
This layer does not require understanding of Delaunay tessellation, icosahedral symmetry, codon residuals, or Kendall tau. It requires only the ability to compare a declared failure condition to an observed result.
Underlying computations, eligibility logic, and statistical controls are documented in Layers 2-4 of the evidence stack.
MH-PRED-V1-WWPDB-P1
PDB-1 — Equilateral-Triad Enrichment in Resolved Cores
VALIDATED
| Prediction | "The distribution of Delaunay contact-triangle equilateralness Q▵ will shift toward 1 in resolved protein cores relative to surface, mobile, or unresolved structural terrain." |
| Falsifier | "No positive core shift or a stable reversed shift." |
| Eligible terrain | 3 entries with sufficient Cα coordinate resolution (9ZYS, 10ZK, 10DT) across 2 cryo-EM and 1 X-ray structure |
| Thresholds tested | 4 quantile cutoffs (Q75, Q80, Q85, Q90) per entry = 12 independent direction tests |
| Observed result | core > surface at all 4 thresholds for all 3 entries. Margins: +0.090 (9ZYS), +0.022 (10ZK), confirmed (10DT) |
| Direction reversals | 0 of 12 |
| Exclusions | Entries without sufficient Cα models, cryo-EM entries below resolution threshold |
| Falsifier triggered? | No - positive core shift present; direction did not reverse |
| Verdict | VALIDATED |
| Qualification | None. Clean coordinate-geometry terrain. No consensus refinement processing between physical reality and the tested quantity. |
MH-PRED-V1-WWPDB-P2
PDB-2 — {2,3,5} Axis Scaffold Recovery in Near-Spherical Cages
VALIDATED
| Prediction | "Eligible closed cages will recover coordinate-derived 2-fold, 3-fold, and 5-fold rotational axes, with the full directional structure approaching 12 vertex, 20 face-normal, and 30 edge-midpoint directions." |
| Falsifier | "A closed eligible cage requires an unrelated axis family." |
| Topology Sentinel | "An icosahedral metadata label is not accepted as the geometric result. The axes must be recovered from coordinates." - Prediction document Section XI.7 |
| Eligible terrain | 9ZYS - Phage Oekolampad (Bas18) icosahedral capsid, cryo-EM 3.7 Å, deposited with I symmetry metadata |
| Observed result | C₂ family: 49 found, 6 clusters, best RMSD 7.69 Å. C₃ family: 16 found, 1 cluster, best RMSD 8.44 Å. C₅ family: 3 found, 1 cluster, best RMSD 9.95 Å. All within RMSD < 10 Å threshold. Symmetry metadata not used. |
| All 3 families present? | Yes |
| Falsifier triggered? | No - all three rotation families {C₂, C₃, C₅} recovered from coordinates without using metadata |
| Verdict | VALIDATED |
| Qualification | 9ZYS is a single asymmetric unit (9 chains) of a 60-ASU icosahedral assembly. Recovering a fraction of the full 15/10/6 axis counts from one ASU is geometrically expected. The requirement is all three families present - they are. |
MH-PRED-V1-WWPDB-P3
PDB-3 — Codon Z-Cascade Residual vs. Structural Resolution
VALIDATED
| Prediction | "The first-shell codon residual D𝑖 = |Δ𝑖 + ∑ⱼ∈N(i) Δⱼ| will be lower in resolved core neighborhoods and higher in mobile or unresolved regions." |
| Falsifier | "No association between D𝑖 and structural resolution, or a reversed association." |
| Eligible terrain | 10DT - Fgr kinase SH3-SH2-linker, X-ray 1.8 Å, 177 residues with recoverable CDS |
| Observed result | D𝑖 core < mobile: True. Pearson r(D𝑖, B-factor) = +0.0445. Permutation control (shuffled Δ): -0.1674. Real r exceeds null by +0.212 in predicted direction. |
| Association present? | Yes - positive, in predicted direction, above permutation null |
| Falsifier triggered? | No - positive association present; no reversed association |
| Verdict | VALIDATED |
| Qualification | Signal magnitude (r = +0.045) is modest. B-factor is a consensus-refined parameter with a long processing chain between substrate and measurement. Signal attenuation is expected and documented in Layer 7. The stated falsifier tests direction, not magnitude. Direction is positive. |
MH-PRED-V1-WWPDB-P4
PDB-4 — Synonymous-Codon Orientation Delta Stratification
VALIDATED
| Prediction | "Within synonymous-codon matched residues, lower locally cancelled codon Orientation Δ will correspond to greater structural order and lower mobility or unresolved-density probability." |
| Falsifier | "No synonymous-codon stratification or a stable reversed direction." |
| Eligible terrain | 129 of 293 entries with recoverable CDS data via UniProt-RefSeq linkage |
| Total pairs evaluated | 1,051,902 synonymous-codon pairwise comparisons across 8,507 residues and 16 amino acid classes |
| Observed result | Concordant C = 544,325. Discordant D = 507,577. Global Kendall tau = +0.034935. Concordant fraction = 0.5175 (null = 0.5000). Aggregate direction across 1,051,902 synonymous-codon pairwise comparisons was positive. |
| Control | Permutation control (B-factors shuffled within AA class, pooled across all 129 entries, seed 42): tau = +0.025924. Real tau exceeds control in predicted direction. |
| Control note | The permutation is amino-acid-preserving but not structure-preserving. It does not establish a cluster-robust p-value. A within-entry block permutation would address the open confidence question - which is separate from the verdict. |
| Direction reversed? | No - global direction positive; no stable reversal |
| Stratification present? | Yes - C exceeds D across 1,051,902 pairs |
| Falsifier triggered? | No - stratification present; direction did not stably reverse |
| Verdict | VALIDATED |
| Qualification | Cluster-robust p-value unresolved from the existing permutation alone. Open confidence question has no authority over the preregistered direction-based falsifier. The falsifier is direction-only. The direction held. |
Full Evidence Stack — wwPDB July 15, 2026
- Layer 1 → Sealed Prediction Record
- Layer 2 → Target-Terrain Manifest
- Layer 3 → Transformation Ledger
- Layer 4 → Executable Scorer
- Layer 5 → Agent Audit Packet
- Layer 6 → This page
- Layer 7 → Evidentiary Narrative
- Layer 8 → Replication Challenge