Foldiff · TnI WT / R145W · Interfaces and local shape · 2026-10-15 SOURCE / ИСТОЧНИК Dryad https://doi.org/10.5061/dryad.rv15dv4cr (CC0), MD_inputs.zip. 288 restrained last-frame PDBs; 2 genotypes × 2 protocols × 24 windows × 3 trials. Every source PDB SHA-256 is rechecked against samples/endpoints.json. Source atom identifiers: segment / residue / atom, not positional array indices. The archive and full-atom PDBs stay local. Browser assets contain derived numbers and existing reduced Cα models only. Derived JSON/CSV are exportable. CONTACT MAP / КАРТА КОНТАКТОВ All heavy atoms (excluding H and D) of TnI B:41–166 versus TnC C and three actin subunits E/G/I. Each residue pair receives the minimum Euclidean atom distance. Distance is computed before any alignment, in deposited whole-complex coordinates. No periodic minimum-image correction is applied without box history. Keep the union of pairs within 6 Å in at least one of all 288 snapshots; retain ALL 288 distances for each retained pair (not clipped, not just positive contacts). The UI restricts rows to pairs within the chosen cutoff in at least one of the six snapshots in the selected window, partner and TnI region. No row means no pair passes that filter, not an unavailable measurement. Values in JSON: 4 decimal places; UI: 2. Contact decisions use stored 4-place values with distance <= cutoff. Within 0.00005 Å of a cutoff, rounding may affect membership; thresholds themselves are analytical choices and do not define a chemical bond or interaction energy. Thresholds 4 / 4.5 / 5 / 6 Å are sensitivity settings, not literature-fitted cutoffs. ARG and TRP have different atom sets; side-chain size can affect minimum distances. 3D highlights Cα locations, NOT the closest heavy-atom pair. Numbers use full atoms. LOCAL SHAPE / ЛОКАЛЬНАЯ ФОРМА For each of nine WT × R145W pairs per window, fit using Kabsch with proper rotation: 1. TnC Cα C:2–161; per-residue displacement of TnI Cα B:41–166. 2. Actin G Cα 1–375; the same per-residue displacement. 3. TnI Cα B:41–166; removes rigid motion of this TnI fragment. 4. Independent nine-Cα local fits: window centered at residue, shifted inwards at termini to retain nine atoms. Plot local RMSD, not displacement of its center. A residue can move relative to a partner while preserving local backbone shape. This is not RMSF, side-chain analysis, secondary-structure assignment or an energy. The existing 3D viewer always uses the common TnC reference, regardless of selected analysis anchor. Its caption states this. Local-fit numbers do not recolor/move 3D. REPEATABILITY / ПОВТОРЯЕМОСТЬ Per window: three WT and three mutant final snapshots. Trial numbers are not matched. Contact count n/3 means n final snapshots, NOT time occupancy or population fraction. Report minimum/median/maximum distances and 3/3-versus-0/3 contrasts at four cutoffs. Counts across 24 windows are counts of biased configurations, NOT probabilities. Do not merge HP and SP protocols or treat window index as elapsed time. Shape: nine cross pairs, three within-WT pairs, three within-mutant pairs. Display their medians and min/max bands separately. Nine cross pairs share six snapshots; not nine independent experimental replicates. No p-value, confidence interval, significance, causal/clinical interpretation or bootstrap is produced. Within-group ranges provide context, not a formal null distribution. LIMITS / ОГРАНИЧЕНИЯ SP configurations specify WT 10 ns, mutant 25 ns; actual continuations and endpoint ages remain unknown. No trajectory, PMF, contact occupancy or kinetics reconstructed. Chain continuity checks do not prove correct interchain periodic imaging. These are restrained simulation snapshots, not 288 independent wet-lab measurements. The full trajectory / CV time series is still required for energetic reconstruction. VALIDATION / ПРОВЕРКИ Eight automated test groups in test_interfaces.py: - full inventory, finite distances and source manifest hashes; - closest-heavy contacts independently checked with MDTraj on four full-atom files; - local RMSD independently checked with MDTraj's QCP method; - rigid rotation/translation invariance; isolated TnI translation vs local shape; - controlled local deformation is detected and remote region stays unchanged; - cutoff boundaries and overlap handling; - cross-pair inventory and median-band agreement; - held-out R145W SP/window17/trial3 full-atom mapping and agreement with previous 145–G334 distances across all 288 snapshots. These validate arithmetic and mapping, not biological or clinical predictive accuracy. METHOD REFERENCES https://mdtraj.readthedocs.io/en/stable/api/generated/mdtraj.compute_contacts.html https://mdtraj.readthedocs.io/en/stable/api/generated/mdtraj.rmsd.html https://doi.org/10.1021/acs.jcim.3c00388 REPRODUCE Use the project scientific Python environment (NumPy, SciPy, MDTraj). python analyze_interfaces.py /path/to/doi_10_5061_dryad_rv15dv4cr__v20230512.zip python -m unittest test_interfaces -v Outputs: samples/interfaces.json, interface-contacts.csv, local-shape.csv. JSON includes source manifest, source archive and analysis-code SHA-256.